SlideShare uma empresa Scribd logo
1 de 7
Baixar para ler offline
I5 geetha4 suraiya
Special Indices for LaaLaLaa Lyric Analysis
                                &
                     Generation Framework
           Suriyah M, Madhan Karky, T V Geetha, & Ranjani Parthasarathi
                            {suriyah.cse@gmail.com, madhankarky@gmail.com,
                                   tv_g@hotmail.com, rp@annauniv.edux}
                                      Tamil Computing Lab (TaCoLa),
                        College of Engineering Guindy, Anna University, Chennai.

Abstract
With the advent of computational tools for creativity, it becomes inevitable to design data structures
which cater to the specific needs of the creative form considered. A lyric generator has to retrieve
words fast based on Part of Speech and rhyme. This paper aims at building special indices for the
LaaLaLaa Lyric Generator framework based on POS and rhyme to facilitate faster retrieval. The
retrieval times of the proposed model and the conservative word indexed model are compared. The
indexing is based on the KNM(Kuril, Nedil, Mei) pattern and the letters that occur in the rhyming
spots of the words. The data structure is organized as hash tables to ensure best retrieval complexity.
Separate hash tables for each POS and rhyming scheme are created and populated. Here, the key
would be the meter pattern with the letters occurring at the rhyming spots and the value would be the
list of all those words which fall under the key’s constraint. When the word indexed and meter rhyme
indexed retrievals were compared, the latter reduced the average retrieval time drastically. There
were not steep variations in the retrieval times as was in the former approach. This remarkable
efficiency was traded-off with space.

1. Introduction
Tamil, one of the oldest languages, has a very rich literary history dating back to two thousand years.
We have more than two thousand lyrics being written in this language in the form of film songs,
advertisements, jingles, private albums etc. With the advent of computational tools for creativity, it
becomes inevitable to design data structures which cater to the specific needs of the creative form
considered. Tools for poem generation, story generation and lyric generation have been proposed.

A poem has to have three qualities – meaningfulness, poeticness and grammaticality [1]. A lyric is a
poem which has constraints of having to satisfy a tune and a theme. This paper talks about building
special indices for the LaaLaLaa lyric generator framework[2] to aid faster retrieval of words based on
POS and rhyme.

A lyric generator would require the retrieval of words of a particular meter and particular letters at
rhyming spots. This would maximize the poeticness of the lyric generated with the increase in
rhymes. This retrieval process, when carried out on an un-indexed word database, is too expensive
as it would take separate processing for meter, and each of the three rhymes in Tamil. To facilitate
faster retrieval of words satisfying these constraints, the word database has to be indexed based on




                                                 287
meter-pattern and rhyme. This makes indexing of the word database based on the abovesaid
constraints necessary.

This paper is organized as eight sections. The second section discusses about a few existing works in
this area. The third section gives an overview of the Rhyme schemes in Tamil. An overall view of the
system is given by the fourth section while the fifth section talks about the approach proposed for
indexing. This work concludes with the results obtained and scope for future work in this area.

2. Background
Though significant number of works has been done in the arena of poetry generation in other
languages, there is only less number in Tamil.

The “Automatic Generation of Tamil Lyrics for Melodies” [3] identifies the required syllable pattern
for the lyric and passes this to a sentence generation module which generates meaningful phrases that
match the pattern. This system generates rhyme based on maximum substring match and fails to
make use of the three rhyming schemes that are specific to Tamil language. “LaaLaLaa - A Tamil Lyric
Analysis and Generation Framework” [2] generates Tamil lyrics for POS tagged pattern with words
from a rhyme finder according to rhyming schemes in Tamil. Nichols et al[5] investigate the
assumption that songwriters tend to align low-level features of a song’s text with musical features. K.
Narayana Murthy[4] suggests having a non-dense TRIE index in main memory and a dense index file
stored in secondary memory.

Anna Babarczy et al[6] suggested a hypothesis that a metaphoric sentence should include both source-
domain and target-domain expressions. This assumption was tested relying on three different
methods of selecting target-domain and source-domain expressions: a psycholinguistic word
association method, a dictionary method and a corpus-based method. Hu, Downie and Ehmann[7]
examine the role lyric text can play in improving audio music mood classification. Mahedero et al[8]
argue that a textual analysis of a song can generate ground truth data that can be used to validate
results from purely acoustic methods. Mayer et al[9] present a novel set of features developed for
textual analysis of song lyrics, and combine them with and compare them to classical bag-of-words
indexing approaches and results for musical genre classification on a test collection in order to
demonstrate our analysis. “Semantic analysis of song lyrics” studies the use of song lyrics for
automatic indexing of music. Netzer et al explore the usage of Word Association Norms (WANs) as
an alternative lexical knowledge source to analyze linguistic computational creativity. Logan et al. use
song lyrics for tracks by 399 artists to determine artist similarity[12].

3. Rhyme Schemes and Rhyme Patterns
Rhyme Schemes: English has number of rhyme effects like assonance, consonance, perfect, imperfect,
masculine, feminine etc. This arises from the variation in stress patterns of words, lack of clear cut
description about the spots where rhymes can occur.

In Tamil, the grapheme and phoneme are bound stronger than in English. There are 3 characteristic
rhyme schemes in Tamil – Monai (ேமாைன), Edhugai (எ ைக) and Iyaibu (இைய ).

Two words are said to rhyme in monai if their first letters are the same, in edhugai if their second
letters are the same and in iyaibu if their last letters are the same.

                                                     288
Examples: பறைவ and ப ைச rhyme in monai as they start with the same letter.

அ     வ and வ                 rhyme in edhugai as they share the same second letter.

யா ைக and வா                      ைக rhyme in iyaibu as they share the same last letter.

            As one may infer, two words can rhyme in more than one pattern also.



Examples: அ          வ and                வ rhyme in edhugai and iyaibu.

கவ ைதக           and கவ ஞ க                       rhyme in all the three schemes.

Meter Pattern: One way of classifying alphabets of the Tamil language is based on the time interval
(மா திைர - maathirai) for which they are pronounced. One maathirai corresponds to the time taken
to wink the eyelid. The types of letters in this classification are

Nedil (N) (ெந           ) - Those alphabets which are pronounced for the time interval of 2 maathirai.

Kuril (K) (     றி    ) - Alphabets which take 1 maathirai to be pronounced.

Mei (M) (ெம          ) - Alphabets which are pronounced for 0.5 maathirai.

Meter pattern of a word refers to its Kuril Nedil Mei pattern.

For example, the meter pattern of the word பாட                        is NKM as பா is a Nedil(N), ட is a Kuril(K) and
    is a Mei(M).

The indexing methodology proposed needs to take care of both meter pattern and rhyme to facilitate
faster retrieval of words for the LaaLaLaa Lyric Generation and Analysis framework[2].

4. Overview of the System

                                   Word object
     Word                           convertor
      DB
                       Rhyme                        Meter
                      extractor                     Pattern
                                                   Extractor
                                  Index Builder




                      Rhyme Meter
                      Index



Figure 1. Overview of the system.

Each word from the word database is converted to an object. Meter pattern and the alphabets at the
Rhyming positions of each word are found out by Meter Pattern Extractor and Rhyme Extractor
respectively. Using the data obtained, the Index builder builds the Rhyme Meter Index aiding faster
retrieval than the normal un-indexed retrieval.




                                                                  289
5. Indexing Algorithm

                                                Part of Speech
                                                             Letter1        Words
                                              MeterPattern1
                                                             Letter2        Words
                                ேமாைன
                                                             Letter1        Words
                                              MeterPattern2
                                                             Letter2        Words
                                                             Letter1        Words
                                              MeterPattern1
                                                             Letter2        Words
                                 எ     ைக
                                                             Letter1        Words
                                              MeterPattern2
                                                             Letter2        Words
                                                             Letter1        Words
                                              MeterPattern1
                                                             Letter2        Words
                                 இைய
                                                             Letter1        Words
                                              MeterPattern2
                                                             Letter2        Words
                                             Figure 2. Indexing Logic

This indexing has been designed to facilitate fast retrieval of words specifically for lyric generation.
For instance, the system would need a word of a particular meter pattern with a particular letter
rhyming in monai scheme.

The data structure is organized as hash tables with separate tables for each Part Of Speech and
Rhyming Scheme. For example, there is a separate table for Nouns’ monai, Nouns’ edhugai, Nouns’
iyaibu and so on. Here, the keys are the Meter Pattern and the Letter at the particular Rhyming spot.
The values are the words corresponding to the particular Meter pattern and letter.
The algorithm : The word database is scanned word by word and is indexed based on meter pattern
and rhyme. Here,
α ← number of words in the word database;
wi← ith word in the word database;
β ← meter pattern of wi ;
monaiKey ← key of wi corresponding to monai;
edhugaiKey ← key of wi corresponding to edhugai;
iyaybuKey ← lkey of wi corresponding to iyaybu;
begin
     i = 1, 2, 3 …α do

     β ← MeterPattern of        w i;
     monaiKey ← firstLetter ( wi) + β ;
     edhugaiKey ← secondLetter (wi) + β;
     iyaybuKey ← lastLetter (wi) + β ;
     Add the word to the list of words with the   respective keys in the respective Hashtables.
end;




                                                       290
For each word in the word database, meter pattern is extracted first followed by letters at the rhyming
spots namely, first, second and last positions (for monai, edhugai and iyaibu respectively). Keys for
monai, edhugai and iyaybu are found out using the abovesaid equations. The word is added to the
monai, edhugai and iyaibu hashtables with keys monaiKey, edhugaiKey and iyaybuKey respectively
if those keys don’t appear previously in the tables. Else, the word is added to the list of words with
that key.

Hash-tables are chosen for the implementation as they have the retrieval complexity of O(1).

6. Results




                          Figure 3. Word Indexed Vs Meter Rhyme Indexed Approach


Indexing has brought about a drastic increase in the speed of retrieval of the words rhyming with a
given word. Using the Word Indexed approach, the time complexity was O(α) where α was the total
number of words. But after Meter Rhyme indexing, the complexity has become O(1) due to the use of
hash table which is a very efficient data structure for retrieval. Rhyming words for a sample of 500
words were retrieved using both the approaches and the above mentioned graph was obtained. The
Word indexed system tooK 875.47millisecond in an average while the Meter Indexed system tooK
1.90millisecond only.From the graph it can also be inferred that there are steep variations in the Word-
Indexed approach while the Meter-Rhyme Indexed approach does not show such steep variations and
is consistent. In terms of time efficiency, Meter-Rhyme indexed approach is evidently superior
compared to Word-Indexed approach. In terms of space, it is not so efficient as each word will occur
not once, but nine times in various Hash tables.

7. References
        Hisar Maruli Manurung: “An evolutionary algorithm approach to poetry generation”, Thesis
        for Doctor of Philosophy, University of Edinburgh, 2003.

        Sowmiya Dharmalingam., Madhan KarKy. “LaaLaLaa - A Tamil Lyric Analysis and
        Generation FrameworK” in World Classical Tamil Conference – June 2010, Coimbatore




                                                   291
RamaKrishnan, A., S. Kuppan, and S.L. Devi. “Automatic Generation of Tamil Lyrics for
Melodies”     in   NAACL HLT WorKshop on Computational Approaches to Linguistic
Creativity. 2009. Colorado.

K. Narayana Murthy "An Indexing Technique for Efficient Retrieval from Large Dictionaries",
National     Conference   on   Information   Technology   NCIT-97,   21-23   December   1997,
Bhubaneswar.

Eric Nichols, Dan Morris, Sumit Basu, Christopher Raphael, “Relationships between lyrics
and melody in popular music”, ISMIR 2009, October 2009, Japan.

Anna Babarczy, IldiKó Bencze, István FeKete, Eszter Simon, “The Automatic Identification of
Conceptual Metaphors in Hungarian Texts: A Corpus-Based Analysis”, Proceedings of The
seventh international conference on Language Resources and Evaluation (LREC), 2010, Malta.

Xiao Hu, J. Stephen Downie, Andreas F. Ehmann, “Lyric Text Mining in Music Mood
Classification”, ISMIR 2009, Japan.

Jose P. G. Mahedero, Alvaro Martınez, Pedro Cano, “Natural Language Processing of Lyrics”,
Proceedings of the 13th annual ACM internationalconference on Multimedia, New YorK, NY,
USA, 2005.

Rudolf Mayer, Robert Neumayer, Andreas Rauber, “Rhyme and style features for musical
genre classification by lyrics”, Proceedings of the 9th International Conference on Music
Information Retrieval (ISMIR’08), Philadelphia, PA, USA, September 14-18, 2008.

Beth Logan, Andrew KositsKy, Pedro Moreno, “Semantic analysis of Song Lyrics”, IEEE
International Conference on Multimedia and Expo (ICME), June 2004.

Yael Netzer, David Gabay,Yoav Goldberg, Michael Elhadad, “GaiKu : Generating HaiKu
with Word Associations Norms”, Workshop on Computational Approaches to Linguistic
Creativity, CALC-2009 in conjunction with NAACL-HLT 2009, Boulder, Colorado.

B. Logan, A. KositsKy, and P. Moreno, “Semantic Analysis of Song Lyrics”, in Proc IEEE
ICME, 2004.




                                         292

Mais conteúdo relacionado

Mais procurados

Welcome to International Journal of Engineering Research and Development (IJERD)
Welcome to International Journal of Engineering Research and Development (IJERD)Welcome to International Journal of Engineering Research and Development (IJERD)
Welcome to International Journal of Engineering Research and Development (IJERD)IJERD Editor
 
Development of morphological analyzer for hindi
Development of morphological analyzer for hindiDevelopment of morphological analyzer for hindi
Development of morphological analyzer for hindiMade Artha
 
ENHANCING THE PERFORMANCE OF SENTIMENT ANALYSIS SUPERVISED LEARNING USING SEN...
ENHANCING THE PERFORMANCE OF SENTIMENT ANALYSIS SUPERVISED LEARNING USING SEN...ENHANCING THE PERFORMANCE OF SENTIMENT ANALYSIS SUPERVISED LEARNING USING SEN...
ENHANCING THE PERFORMANCE OF SENTIMENT ANALYSIS SUPERVISED LEARNING USING SEN...csandit
 
Ijartes v1-i1-002
Ijartes v1-i1-002Ijartes v1-i1-002
Ijartes v1-i1-002IJARTES
 
DESIGN OF A RULE BASED HINDI LEMMATIZER
DESIGN OF A RULE BASED HINDI LEMMATIZERDESIGN OF A RULE BASED HINDI LEMMATIZER
DESIGN OF A RULE BASED HINDI LEMMATIZERcsandit
 
Design of a rule based hindi lemmatizer
Design of a rule based hindi lemmatizerDesign of a rule based hindi lemmatizer
Design of a rule based hindi lemmatizercsandit
 
Usage of regular expressions in nlp
Usage of regular expressions in nlpUsage of regular expressions in nlp
Usage of regular expressions in nlpeSAT Journals
 
Spell checker using Natural language processing
Spell checker using Natural language processing Spell checker using Natural language processing
Spell checker using Natural language processing Sandeep Wakchaure
 
Verb based manipuri sentiment analysis
Verb based manipuri sentiment analysisVerb based manipuri sentiment analysis
Verb based manipuri sentiment analysisijnlc
 
Dependency Analysis of Abstract Universal Structures in Korean and English
Dependency Analysis of Abstract Universal Structures in Korean and EnglishDependency Analysis of Abstract Universal Structures in Korean and English
Dependency Analysis of Abstract Universal Structures in Korean and EnglishJinho Choi
 
Sanskrit in Natural Language Processing
Sanskrit in Natural Language ProcessingSanskrit in Natural Language Processing
Sanskrit in Natural Language ProcessingHitesh Joshi
 
An implementation of apertium based assamese morphological analyzer
An implementation of apertium based assamese morphological analyzerAn implementation of apertium based assamese morphological analyzer
An implementation of apertium based assamese morphological analyzerijnlc
 
IRJET- A System for Determining Sarcasm in Tweets: Sarcasm Detector
IRJET-  	  A System for Determining Sarcasm in Tweets: Sarcasm DetectorIRJET-  	  A System for Determining Sarcasm in Tweets: Sarcasm Detector
IRJET- A System for Determining Sarcasm in Tweets: Sarcasm DetectorIRJET Journal
 
Implementation Of Syntax Parser For English Language Using Grammar Rules
Implementation Of Syntax Parser For English Language Using Grammar RulesImplementation Of Syntax Parser For English Language Using Grammar Rules
Implementation Of Syntax Parser For English Language Using Grammar RulesIJERA Editor
 
Statistically-Enhanced New Word Identification
Statistically-Enhanced New Word IdentificationStatistically-Enhanced New Word Identification
Statistically-Enhanced New Word IdentificationAndi Wu
 

Mais procurados (19)

Welcome to International Journal of Engineering Research and Development (IJERD)
Welcome to International Journal of Engineering Research and Development (IJERD)Welcome to International Journal of Engineering Research and Development (IJERD)
Welcome to International Journal of Engineering Research and Development (IJERD)
 
Development of morphological analyzer for hindi
Development of morphological analyzer for hindiDevelopment of morphological analyzer for hindi
Development of morphological analyzer for hindi
 
ENHANCING THE PERFORMANCE OF SENTIMENT ANALYSIS SUPERVISED LEARNING USING SEN...
ENHANCING THE PERFORMANCE OF SENTIMENT ANALYSIS SUPERVISED LEARNING USING SEN...ENHANCING THE PERFORMANCE OF SENTIMENT ANALYSIS SUPERVISED LEARNING USING SEN...
ENHANCING THE PERFORMANCE OF SENTIMENT ANALYSIS SUPERVISED LEARNING USING SEN...
 
Ijartes v1-i1-002
Ijartes v1-i1-002Ijartes v1-i1-002
Ijartes v1-i1-002
 
Usage of regular expressions in nlp
Usage of regular expressions in nlpUsage of regular expressions in nlp
Usage of regular expressions in nlp
 
DESIGN OF A RULE BASED HINDI LEMMATIZER
DESIGN OF A RULE BASED HINDI LEMMATIZERDESIGN OF A RULE BASED HINDI LEMMATIZER
DESIGN OF A RULE BASED HINDI LEMMATIZER
 
Design of a rule based hindi lemmatizer
Design of a rule based hindi lemmatizerDesign of a rule based hindi lemmatizer
Design of a rule based hindi lemmatizer
 
Usage of regular expressions in nlp
Usage of regular expressions in nlpUsage of regular expressions in nlp
Usage of regular expressions in nlp
 
Spell checker using Natural language processing
Spell checker using Natural language processing Spell checker using Natural language processing
Spell checker using Natural language processing
 
Nlp
NlpNlp
Nlp
 
Verb based manipuri sentiment analysis
Verb based manipuri sentiment analysisVerb based manipuri sentiment analysis
Verb based manipuri sentiment analysis
 
Dependency Analysis of Abstract Universal Structures in Korean and English
Dependency Analysis of Abstract Universal Structures in Korean and EnglishDependency Analysis of Abstract Universal Structures in Korean and English
Dependency Analysis of Abstract Universal Structures in Korean and English
 
Pxc3898474
Pxc3898474Pxc3898474
Pxc3898474
 
Sanskrit in Natural Language Processing
Sanskrit in Natural Language ProcessingSanskrit in Natural Language Processing
Sanskrit in Natural Language Processing
 
An implementation of apertium based assamese morphological analyzer
An implementation of apertium based assamese morphological analyzerAn implementation of apertium based assamese morphological analyzer
An implementation of apertium based assamese morphological analyzer
 
IRJET- A System for Determining Sarcasm in Tweets: Sarcasm Detector
IRJET-  	  A System for Determining Sarcasm in Tweets: Sarcasm DetectorIRJET-  	  A System for Determining Sarcasm in Tweets: Sarcasm Detector
IRJET- A System for Determining Sarcasm in Tweets: Sarcasm Detector
 
Implementation Of Syntax Parser For English Language Using Grammar Rules
Implementation Of Syntax Parser For English Language Using Grammar RulesImplementation Of Syntax Parser For English Language Using Grammar Rules
Implementation Of Syntax Parser For English Language Using Grammar Rules
 
mtp_stageII_final
mtp_stageII_finalmtp_stageII_final
mtp_stageII_final
 
Statistically-Enhanced New Word Identification
Statistically-Enhanced New Word IdentificationStatistically-Enhanced New Word Identification
Statistically-Enhanced New Word Identification
 

Semelhante a I5 geetha4 suraiya

MYANMAR WORDS SORTING
MYANMAR WORDS SORTING MYANMAR WORDS SORTING
MYANMAR WORDS SORTING kevig
 
MYANMAR WORDS SORTING
MYANMAR WORDS SORTING MYANMAR WORDS SORTING
MYANMAR WORDS SORTING kevig
 
Ijarcet vol-2-issue-2-323-329
Ijarcet vol-2-issue-2-323-329Ijarcet vol-2-issue-2-323-329
Ijarcet vol-2-issue-2-323-329Editor IJARCET
 
Duration for Classification and Regression Treefor Marathi Textto- Speech Syn...
Duration for Classification and Regression Treefor Marathi Textto- Speech Syn...Duration for Classification and Regression Treefor Marathi Textto- Speech Syn...
Duration for Classification and Regression Treefor Marathi Textto- Speech Syn...IJERA Editor
 
A COMPARATIVE STUDY OF ROOT-BASED AND STEM-BASED APPROACHES FOR MEASURING THE...
A COMPARATIVE STUDY OF ROOT-BASED AND STEM-BASED APPROACHES FOR MEASURING THE...A COMPARATIVE STUDY OF ROOT-BASED AND STEM-BASED APPROACHES FOR MEASURING THE...
A COMPARATIVE STUDY OF ROOT-BASED AND STEM-BASED APPROACHES FOR MEASURING THE...acijjournal
 
Rule-Based Phonetic Matching Approach for Hindi and Marathi
Rule-Based Phonetic Matching Approach for Hindi and MarathiRule-Based Phonetic Matching Approach for Hindi and Marathi
Rule-Based Phonetic Matching Approach for Hindi and MarathiCSEIJJournal
 
HMM BASED POS TAGGER FOR HINDI
HMM BASED POS TAGGER FOR HINDIHMM BASED POS TAGGER FOR HINDI
HMM BASED POS TAGGER FOR HINDIcscpconf
 
A Combined Approach to Part-of-Speech Tagging Using Features Extraction and H...
A Combined Approach to Part-of-Speech Tagging Using Features Extraction and H...A Combined Approach to Part-of-Speech Tagging Using Features Extraction and H...
A Combined Approach to Part-of-Speech Tagging Using Features Extraction and H...Editor IJARCET
 
Paper id 25201466
Paper id 25201466Paper id 25201466
Paper id 25201466IJRAT
 
A comparative analysis of particle swarm optimization and k means algorithm f...
A comparative analysis of particle swarm optimization and k means algorithm f...A comparative analysis of particle swarm optimization and k means algorithm f...
A comparative analysis of particle swarm optimization and k means algorithm f...ijnlc
 
Phonetic Recognition In Words For Persian Text To Speech Systems
Phonetic Recognition In Words For Persian Text To Speech SystemsPhonetic Recognition In Words For Persian Text To Speech Systems
Phonetic Recognition In Words For Persian Text To Speech Systemspaperpublications3
 
Shallow parser for hindi language with an input from a transliterator
Shallow parser for hindi language with an input from a transliteratorShallow parser for hindi language with an input from a transliterator
Shallow parser for hindi language with an input from a transliteratorShashank Shisodia
 
Developing links of compound sentences for parsing through marathi link gramm...
Developing links of compound sentences for parsing through marathi link gramm...Developing links of compound sentences for parsing through marathi link gramm...
Developing links of compound sentences for parsing through marathi link gramm...ijnlc
 
5a use of annotated corpus
5a use of annotated corpus5a use of annotated corpus
5a use of annotated corpusThennarasuSakkan
 
Implementation of Text To Speech for Marathi Language Using Transcriptions Co...
Implementation of Text To Speech for Marathi Language Using Transcriptions Co...Implementation of Text To Speech for Marathi Language Using Transcriptions Co...
Implementation of Text To Speech for Marathi Language Using Transcriptions Co...IJERA Editor
 
Identification of Rhetorical Structure Relation from Discourse Marker in Beng...
Identification of Rhetorical Structure Relation from Discourse Marker in Beng...Identification of Rhetorical Structure Relation from Discourse Marker in Beng...
Identification of Rhetorical Structure Relation from Discourse Marker in Beng...idescitation
 

Semelhante a I5 geetha4 suraiya (20)

MYANMAR WORDS SORTING
MYANMAR WORDS SORTING MYANMAR WORDS SORTING
MYANMAR WORDS SORTING
 
MYANMAR WORDS SORTING
MYANMAR WORDS SORTING MYANMAR WORDS SORTING
MYANMAR WORDS SORTING
 
FIRE2014_IIT-P
FIRE2014_IIT-PFIRE2014_IIT-P
FIRE2014_IIT-P
 
Ijarcet vol-2-issue-2-323-329
Ijarcet vol-2-issue-2-323-329Ijarcet vol-2-issue-2-323-329
Ijarcet vol-2-issue-2-323-329
 
D3 dhanalakshmi
D3 dhanalakshmiD3 dhanalakshmi
D3 dhanalakshmi
 
Duration for Classification and Regression Treefor Marathi Textto- Speech Syn...
Duration for Classification and Regression Treefor Marathi Textto- Speech Syn...Duration for Classification and Regression Treefor Marathi Textto- Speech Syn...
Duration for Classification and Regression Treefor Marathi Textto- Speech Syn...
 
A COMPARATIVE STUDY OF ROOT-BASED AND STEM-BASED APPROACHES FOR MEASURING THE...
A COMPARATIVE STUDY OF ROOT-BASED AND STEM-BASED APPROACHES FOR MEASURING THE...A COMPARATIVE STUDY OF ROOT-BASED AND STEM-BASED APPROACHES FOR MEASURING THE...
A COMPARATIVE STUDY OF ROOT-BASED AND STEM-BASED APPROACHES FOR MEASURING THE...
 
Rule-Based Phonetic Matching Approach for Hindi and Marathi
Rule-Based Phonetic Matching Approach for Hindi and MarathiRule-Based Phonetic Matching Approach for Hindi and Marathi
Rule-Based Phonetic Matching Approach for Hindi and Marathi
 
HMM BASED POS TAGGER FOR HINDI
HMM BASED POS TAGGER FOR HINDIHMM BASED POS TAGGER FOR HINDI
HMM BASED POS TAGGER FOR HINDI
 
Ey4301913917
Ey4301913917Ey4301913917
Ey4301913917
 
A Combined Approach to Part-of-Speech Tagging Using Features Extraction and H...
A Combined Approach to Part-of-Speech Tagging Using Features Extraction and H...A Combined Approach to Part-of-Speech Tagging Using Features Extraction and H...
A Combined Approach to Part-of-Speech Tagging Using Features Extraction and H...
 
Paper id 25201466
Paper id 25201466Paper id 25201466
Paper id 25201466
 
A comparative analysis of particle swarm optimization and k means algorithm f...
A comparative analysis of particle swarm optimization and k means algorithm f...A comparative analysis of particle swarm optimization and k means algorithm f...
A comparative analysis of particle swarm optimization and k means algorithm f...
 
Phonetic Recognition In Words For Persian Text To Speech Systems
Phonetic Recognition In Words For Persian Text To Speech SystemsPhonetic Recognition In Words For Persian Text To Speech Systems
Phonetic Recognition In Words For Persian Text To Speech Systems
 
Shallow parser for hindi language with an input from a transliterator
Shallow parser for hindi language with an input from a transliteratorShallow parser for hindi language with an input from a transliterator
Shallow parser for hindi language with an input from a transliterator
 
Developing links of compound sentences for parsing through marathi link gramm...
Developing links of compound sentences for parsing through marathi link gramm...Developing links of compound sentences for parsing through marathi link gramm...
Developing links of compound sentences for parsing through marathi link gramm...
 
5a use of annotated corpus
5a use of annotated corpus5a use of annotated corpus
5a use of annotated corpus
 
Implementation of Text To Speech for Marathi Language Using Transcriptions Co...
Implementation of Text To Speech for Marathi Language Using Transcriptions Co...Implementation of Text To Speech for Marathi Language Using Transcriptions Co...
Implementation of Text To Speech for Marathi Language Using Transcriptions Co...
 
Identification of Rhetorical Structure Relation from Discourse Marker in Beng...
Identification of Rhetorical Structure Relation from Discourse Marker in Beng...Identification of Rhetorical Structure Relation from Discourse Marker in Beng...
Identification of Rhetorical Structure Relation from Discourse Marker in Beng...
 
I1 geetha3 revathi
I1 geetha3 revathiI1 geetha3 revathi
I1 geetha3 revathi
 

Mais de Jasline Presilda (20)

I6 mala3 sowmya
I6 mala3 sowmyaI6 mala3 sowmya
I6 mala3 sowmya
 
I4 madankarky3 subalalitha
I4 madankarky3 subalalithaI4 madankarky3 subalalitha
I4 madankarky3 subalalitha
 
I2 madankarky1 jharibabu
I2 madankarky1 jharibabuI2 madankarky1 jharibabu
I2 madankarky1 jharibabu
 
Hari tamil-complete details
Hari tamil-complete detailsHari tamil-complete details
Hari tamil-complete details
 
H4 neelavathy
H4 neelavathyH4 neelavathy
H4 neelavathy
 
H3 anuraj
H3 anurajH3 anuraj
H3 anuraj
 
H2 gunasekaran
H2 gunasekaranH2 gunasekaran
H2 gunasekaran
 
H1 iniya nehru
H1 iniya nehruH1 iniya nehru
H1 iniya nehru
 
G3 chandrakala
G3 chandrakalaG3 chandrakala
G3 chandrakala
 
G2 selvakumar
G2 selvakumarG2 selvakumar
G2 selvakumar
 
G1 nmurugaiyan
G1 nmurugaiyanG1 nmurugaiyan
G1 nmurugaiyan
 
Front matter
Front matterFront matter
Front matter
 
F2 pvairam sarathy
F2 pvairam sarathyF2 pvairam sarathy
F2 pvairam sarathy
 
F1 ferdinjoe
F1 ferdinjoeF1 ferdinjoe
F1 ferdinjoe
 
Emerging
EmergingEmerging
Emerging
 
E3 ilangkumaran
E3 ilangkumaranE3 ilangkumaran
E3 ilangkumaran
 
E2 tamilselvan
E2 tamilselvanE2 tamilselvan
E2 tamilselvan
 
E1 geetha2 karthikeyan
E1 geetha2 karthikeyanE1 geetha2 karthikeyan
E1 geetha2 karthikeyan
 
D5 radha chellappan
D5 radha chellappanD5 radha chellappan
D5 radha chellappan
 
D4 sundaram
D4 sundaramD4 sundaram
D4 sundaram
 

Último

BBA 205 BUSINESS ENVIRONMENT UNIT I.pptx
BBA 205 BUSINESS ENVIRONMENT UNIT I.pptxBBA 205 BUSINESS ENVIRONMENT UNIT I.pptx
BBA 205 BUSINESS ENVIRONMENT UNIT I.pptxProf. Kanchan Kumari
 
DNA and RNA , Structure, Functions, Types, difference, Similarities, Protein ...
DNA and RNA , Structure, Functions, Types, difference, Similarities, Protein ...DNA and RNA , Structure, Functions, Types, difference, Similarities, Protein ...
DNA and RNA , Structure, Functions, Types, difference, Similarities, Protein ...AKSHAYMAGAR17
 
BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (FRIE...
BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (FRIE...BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (FRIE...
BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (FRIE...Nguyen Thanh Tu Collection
 
BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (GLOB...
BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (GLOB...BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (GLOB...
BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (GLOB...Nguyen Thanh Tu Collection
 
VIT336 – Recommender System - Unit 3.pdf
VIT336 – Recommender System - Unit 3.pdfVIT336 – Recommender System - Unit 3.pdf
VIT336 – Recommender System - Unit 3.pdfArthyR3
 
The First National K12 TUG March 6 2024.pdf
The First National K12 TUG March 6 2024.pdfThe First National K12 TUG March 6 2024.pdf
The First National K12 TUG March 6 2024.pdfdogden2
 
POST ENCEPHALITIS case study Jitendra bhargav
POST ENCEPHALITIS case study  Jitendra bhargavPOST ENCEPHALITIS case study  Jitendra bhargav
POST ENCEPHALITIS case study Jitendra bhargavJitendra Bhargav
 
25 CHUYÊN ĐỀ ÔN THI TỐT NGHIỆP THPT 2023 – BÀI TẬP PHÁT TRIỂN TỪ ĐỀ MINH HỌA...
25 CHUYÊN ĐỀ ÔN THI TỐT NGHIỆP THPT 2023 – BÀI TẬP PHÁT TRIỂN TỪ ĐỀ MINH HỌA...25 CHUYÊN ĐỀ ÔN THI TỐT NGHIỆP THPT 2023 – BÀI TẬP PHÁT TRIỂN TỪ ĐỀ MINH HỌA...
25 CHUYÊN ĐỀ ÔN THI TỐT NGHIỆP THPT 2023 – BÀI TẬP PHÁT TRIỂN TỪ ĐỀ MINH HỌA...Nguyen Thanh Tu Collection
 
Riti theory by Vamana Indian poetics.pptx
Riti theory by Vamana Indian poetics.pptxRiti theory by Vamana Indian poetics.pptx
Riti theory by Vamana Indian poetics.pptxDhatriParmar
 
PHARMACOGNOSY CHAPTER NO 5 CARMINATIVES AND G.pdf
PHARMACOGNOSY CHAPTER NO 5 CARMINATIVES AND G.pdfPHARMACOGNOSY CHAPTER NO 5 CARMINATIVES AND G.pdf
PHARMACOGNOSY CHAPTER NO 5 CARMINATIVES AND G.pdfSumit Tiwari
 
Dhavni Theory by Anandvardhana Indian Poetics
Dhavni Theory by Anandvardhana Indian PoeticsDhavni Theory by Anandvardhana Indian Poetics
Dhavni Theory by Anandvardhana Indian PoeticsDhatriParmar
 
AUDIENCE THEORY - PARTICIPATORY - JENKINS.pptx
AUDIENCE THEORY - PARTICIPATORY - JENKINS.pptxAUDIENCE THEORY - PARTICIPATORY - JENKINS.pptx
AUDIENCE THEORY - PARTICIPATORY - JENKINS.pptxiammrhaywood
 
Quantitative research methodology and survey design
Quantitative research methodology and survey designQuantitative research methodology and survey design
Quantitative research methodology and survey designBalelaBoru
 
3.12.24 The Social Construction of Gender.pptx
3.12.24 The Social Construction of Gender.pptx3.12.24 The Social Construction of Gender.pptx
3.12.24 The Social Construction of Gender.pptxmary850239
 
BBA 205 BE UNIT 2 economic systems prof dr kanchan.pptx
BBA 205 BE UNIT 2 economic systems prof dr kanchan.pptxBBA 205 BE UNIT 2 economic systems prof dr kanchan.pptx
BBA 205 BE UNIT 2 economic systems prof dr kanchan.pptxProf. Kanchan Kumari
 
LEAD6001 - Introduction to Advanced Stud
LEAD6001 - Introduction to Advanced StudLEAD6001 - Introduction to Advanced Stud
LEAD6001 - Introduction to Advanced StudDr. Bruce A. Johnson
 
Pharmacology chapter No 7 full notes.pdf
Pharmacology chapter No 7 full notes.pdfPharmacology chapter No 7 full notes.pdf
Pharmacology chapter No 7 full notes.pdfSumit Tiwari
 
3.14.24 Gender Discrimination and Gender Inequity.pptx
3.14.24 Gender Discrimination and Gender Inequity.pptx3.14.24 Gender Discrimination and Gender Inequity.pptx
3.14.24 Gender Discrimination and Gender Inequity.pptxmary850239
 
LEAD5623 The Economics of Community Coll
LEAD5623 The Economics of Community CollLEAD5623 The Economics of Community Coll
LEAD5623 The Economics of Community CollDr. Bruce A. Johnson
 

Último (20)

BBA 205 BUSINESS ENVIRONMENT UNIT I.pptx
BBA 205 BUSINESS ENVIRONMENT UNIT I.pptxBBA 205 BUSINESS ENVIRONMENT UNIT I.pptx
BBA 205 BUSINESS ENVIRONMENT UNIT I.pptx
 
DNA and RNA , Structure, Functions, Types, difference, Similarities, Protein ...
DNA and RNA , Structure, Functions, Types, difference, Similarities, Protein ...DNA and RNA , Structure, Functions, Types, difference, Similarities, Protein ...
DNA and RNA , Structure, Functions, Types, difference, Similarities, Protein ...
 
BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (FRIE...
BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (FRIE...BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (FRIE...
BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (FRIE...
 
Problems on Mean,Mode,Median Standard Deviation
Problems on Mean,Mode,Median Standard DeviationProblems on Mean,Mode,Median Standard Deviation
Problems on Mean,Mode,Median Standard Deviation
 
BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (GLOB...
BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (GLOB...BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (GLOB...
BÀI TẬP BỔ TRỢ TIẾNG ANH 11 THEO ĐƠN VỊ BÀI HỌC - CẢ NĂM - CÓ FILE NGHE (GLOB...
 
VIT336 – Recommender System - Unit 3.pdf
VIT336 – Recommender System - Unit 3.pdfVIT336 – Recommender System - Unit 3.pdf
VIT336 – Recommender System - Unit 3.pdf
 
The First National K12 TUG March 6 2024.pdf
The First National K12 TUG March 6 2024.pdfThe First National K12 TUG March 6 2024.pdf
The First National K12 TUG March 6 2024.pdf
 
POST ENCEPHALITIS case study Jitendra bhargav
POST ENCEPHALITIS case study  Jitendra bhargavPOST ENCEPHALITIS case study  Jitendra bhargav
POST ENCEPHALITIS case study Jitendra bhargav
 
25 CHUYÊN ĐỀ ÔN THI TỐT NGHIỆP THPT 2023 – BÀI TẬP PHÁT TRIỂN TỪ ĐỀ MINH HỌA...
25 CHUYÊN ĐỀ ÔN THI TỐT NGHIỆP THPT 2023 – BÀI TẬP PHÁT TRIỂN TỪ ĐỀ MINH HỌA...25 CHUYÊN ĐỀ ÔN THI TỐT NGHIỆP THPT 2023 – BÀI TẬP PHÁT TRIỂN TỪ ĐỀ MINH HỌA...
25 CHUYÊN ĐỀ ÔN THI TỐT NGHIỆP THPT 2023 – BÀI TẬP PHÁT TRIỂN TỪ ĐỀ MINH HỌA...
 
Riti theory by Vamana Indian poetics.pptx
Riti theory by Vamana Indian poetics.pptxRiti theory by Vamana Indian poetics.pptx
Riti theory by Vamana Indian poetics.pptx
 
PHARMACOGNOSY CHAPTER NO 5 CARMINATIVES AND G.pdf
PHARMACOGNOSY CHAPTER NO 5 CARMINATIVES AND G.pdfPHARMACOGNOSY CHAPTER NO 5 CARMINATIVES AND G.pdf
PHARMACOGNOSY CHAPTER NO 5 CARMINATIVES AND G.pdf
 
Dhavni Theory by Anandvardhana Indian Poetics
Dhavni Theory by Anandvardhana Indian PoeticsDhavni Theory by Anandvardhana Indian Poetics
Dhavni Theory by Anandvardhana Indian Poetics
 
AUDIENCE THEORY - PARTICIPATORY - JENKINS.pptx
AUDIENCE THEORY - PARTICIPATORY - JENKINS.pptxAUDIENCE THEORY - PARTICIPATORY - JENKINS.pptx
AUDIENCE THEORY - PARTICIPATORY - JENKINS.pptx
 
Quantitative research methodology and survey design
Quantitative research methodology and survey designQuantitative research methodology and survey design
Quantitative research methodology and survey design
 
3.12.24 The Social Construction of Gender.pptx
3.12.24 The Social Construction of Gender.pptx3.12.24 The Social Construction of Gender.pptx
3.12.24 The Social Construction of Gender.pptx
 
BBA 205 BE UNIT 2 economic systems prof dr kanchan.pptx
BBA 205 BE UNIT 2 economic systems prof dr kanchan.pptxBBA 205 BE UNIT 2 economic systems prof dr kanchan.pptx
BBA 205 BE UNIT 2 economic systems prof dr kanchan.pptx
 
LEAD6001 - Introduction to Advanced Stud
LEAD6001 - Introduction to Advanced StudLEAD6001 - Introduction to Advanced Stud
LEAD6001 - Introduction to Advanced Stud
 
Pharmacology chapter No 7 full notes.pdf
Pharmacology chapter No 7 full notes.pdfPharmacology chapter No 7 full notes.pdf
Pharmacology chapter No 7 full notes.pdf
 
3.14.24 Gender Discrimination and Gender Inequity.pptx
3.14.24 Gender Discrimination and Gender Inequity.pptx3.14.24 Gender Discrimination and Gender Inequity.pptx
3.14.24 Gender Discrimination and Gender Inequity.pptx
 
LEAD5623 The Economics of Community Coll
LEAD5623 The Economics of Community CollLEAD5623 The Economics of Community Coll
LEAD5623 The Economics of Community Coll
 

I5 geetha4 suraiya

  • 2. Special Indices for LaaLaLaa Lyric Analysis & Generation Framework Suriyah M, Madhan Karky, T V Geetha, & Ranjani Parthasarathi {suriyah.cse@gmail.com, madhankarky@gmail.com, tv_g@hotmail.com, rp@annauniv.edux} Tamil Computing Lab (TaCoLa), College of Engineering Guindy, Anna University, Chennai. Abstract With the advent of computational tools for creativity, it becomes inevitable to design data structures which cater to the specific needs of the creative form considered. A lyric generator has to retrieve words fast based on Part of Speech and rhyme. This paper aims at building special indices for the LaaLaLaa Lyric Generator framework based on POS and rhyme to facilitate faster retrieval. The retrieval times of the proposed model and the conservative word indexed model are compared. The indexing is based on the KNM(Kuril, Nedil, Mei) pattern and the letters that occur in the rhyming spots of the words. The data structure is organized as hash tables to ensure best retrieval complexity. Separate hash tables for each POS and rhyming scheme are created and populated. Here, the key would be the meter pattern with the letters occurring at the rhyming spots and the value would be the list of all those words which fall under the key’s constraint. When the word indexed and meter rhyme indexed retrievals were compared, the latter reduced the average retrieval time drastically. There were not steep variations in the retrieval times as was in the former approach. This remarkable efficiency was traded-off with space. 1. Introduction Tamil, one of the oldest languages, has a very rich literary history dating back to two thousand years. We have more than two thousand lyrics being written in this language in the form of film songs, advertisements, jingles, private albums etc. With the advent of computational tools for creativity, it becomes inevitable to design data structures which cater to the specific needs of the creative form considered. Tools for poem generation, story generation and lyric generation have been proposed. A poem has to have three qualities – meaningfulness, poeticness and grammaticality [1]. A lyric is a poem which has constraints of having to satisfy a tune and a theme. This paper talks about building special indices for the LaaLaLaa lyric generator framework[2] to aid faster retrieval of words based on POS and rhyme. A lyric generator would require the retrieval of words of a particular meter and particular letters at rhyming spots. This would maximize the poeticness of the lyric generated with the increase in rhymes. This retrieval process, when carried out on an un-indexed word database, is too expensive as it would take separate processing for meter, and each of the three rhymes in Tamil. To facilitate faster retrieval of words satisfying these constraints, the word database has to be indexed based on 287
  • 3. meter-pattern and rhyme. This makes indexing of the word database based on the abovesaid constraints necessary. This paper is organized as eight sections. The second section discusses about a few existing works in this area. The third section gives an overview of the Rhyme schemes in Tamil. An overall view of the system is given by the fourth section while the fifth section talks about the approach proposed for indexing. This work concludes with the results obtained and scope for future work in this area. 2. Background Though significant number of works has been done in the arena of poetry generation in other languages, there is only less number in Tamil. The “Automatic Generation of Tamil Lyrics for Melodies” [3] identifies the required syllable pattern for the lyric and passes this to a sentence generation module which generates meaningful phrases that match the pattern. This system generates rhyme based on maximum substring match and fails to make use of the three rhyming schemes that are specific to Tamil language. “LaaLaLaa - A Tamil Lyric Analysis and Generation Framework” [2] generates Tamil lyrics for POS tagged pattern with words from a rhyme finder according to rhyming schemes in Tamil. Nichols et al[5] investigate the assumption that songwriters tend to align low-level features of a song’s text with musical features. K. Narayana Murthy[4] suggests having a non-dense TRIE index in main memory and a dense index file stored in secondary memory. Anna Babarczy et al[6] suggested a hypothesis that a metaphoric sentence should include both source- domain and target-domain expressions. This assumption was tested relying on three different methods of selecting target-domain and source-domain expressions: a psycholinguistic word association method, a dictionary method and a corpus-based method. Hu, Downie and Ehmann[7] examine the role lyric text can play in improving audio music mood classification. Mahedero et al[8] argue that a textual analysis of a song can generate ground truth data that can be used to validate results from purely acoustic methods. Mayer et al[9] present a novel set of features developed for textual analysis of song lyrics, and combine them with and compare them to classical bag-of-words indexing approaches and results for musical genre classification on a test collection in order to demonstrate our analysis. “Semantic analysis of song lyrics” studies the use of song lyrics for automatic indexing of music. Netzer et al explore the usage of Word Association Norms (WANs) as an alternative lexical knowledge source to analyze linguistic computational creativity. Logan et al. use song lyrics for tracks by 399 artists to determine artist similarity[12]. 3. Rhyme Schemes and Rhyme Patterns Rhyme Schemes: English has number of rhyme effects like assonance, consonance, perfect, imperfect, masculine, feminine etc. This arises from the variation in stress patterns of words, lack of clear cut description about the spots where rhymes can occur. In Tamil, the grapheme and phoneme are bound stronger than in English. There are 3 characteristic rhyme schemes in Tamil – Monai (ேமாைன), Edhugai (எ ைக) and Iyaibu (இைய ). Two words are said to rhyme in monai if their first letters are the same, in edhugai if their second letters are the same and in iyaibu if their last letters are the same. 288
  • 4. Examples: பறைவ and ப ைச rhyme in monai as they start with the same letter. அ வ and வ rhyme in edhugai as they share the same second letter. யா ைக and வா ைக rhyme in iyaibu as they share the same last letter. As one may infer, two words can rhyme in more than one pattern also. Examples: அ வ and வ rhyme in edhugai and iyaibu. கவ ைதக and கவ ஞ க rhyme in all the three schemes. Meter Pattern: One way of classifying alphabets of the Tamil language is based on the time interval (மா திைர - maathirai) for which they are pronounced. One maathirai corresponds to the time taken to wink the eyelid. The types of letters in this classification are Nedil (N) (ெந ) - Those alphabets which are pronounced for the time interval of 2 maathirai. Kuril (K) ( றி ) - Alphabets which take 1 maathirai to be pronounced. Mei (M) (ெம ) - Alphabets which are pronounced for 0.5 maathirai. Meter pattern of a word refers to its Kuril Nedil Mei pattern. For example, the meter pattern of the word பாட is NKM as பா is a Nedil(N), ட is a Kuril(K) and is a Mei(M). The indexing methodology proposed needs to take care of both meter pattern and rhyme to facilitate faster retrieval of words for the LaaLaLaa Lyric Generation and Analysis framework[2]. 4. Overview of the System Word object Word convertor DB Rhyme Meter extractor Pattern Extractor Index Builder Rhyme Meter Index Figure 1. Overview of the system. Each word from the word database is converted to an object. Meter pattern and the alphabets at the Rhyming positions of each word are found out by Meter Pattern Extractor and Rhyme Extractor respectively. Using the data obtained, the Index builder builds the Rhyme Meter Index aiding faster retrieval than the normal un-indexed retrieval. 289
  • 5. 5. Indexing Algorithm Part of Speech Letter1 Words MeterPattern1 Letter2 Words ேமாைன Letter1 Words MeterPattern2 Letter2 Words Letter1 Words MeterPattern1 Letter2 Words எ ைக Letter1 Words MeterPattern2 Letter2 Words Letter1 Words MeterPattern1 Letter2 Words இைய Letter1 Words MeterPattern2 Letter2 Words Figure 2. Indexing Logic This indexing has been designed to facilitate fast retrieval of words specifically for lyric generation. For instance, the system would need a word of a particular meter pattern with a particular letter rhyming in monai scheme. The data structure is organized as hash tables with separate tables for each Part Of Speech and Rhyming Scheme. For example, there is a separate table for Nouns’ monai, Nouns’ edhugai, Nouns’ iyaibu and so on. Here, the keys are the Meter Pattern and the Letter at the particular Rhyming spot. The values are the words corresponding to the particular Meter pattern and letter. The algorithm : The word database is scanned word by word and is indexed based on meter pattern and rhyme. Here, α ← number of words in the word database; wi← ith word in the word database; β ← meter pattern of wi ; monaiKey ← key of wi corresponding to monai; edhugaiKey ← key of wi corresponding to edhugai; iyaybuKey ← lkey of wi corresponding to iyaybu; begin i = 1, 2, 3 …α do β ← MeterPattern of w i; monaiKey ← firstLetter ( wi) + β ; edhugaiKey ← secondLetter (wi) + β; iyaybuKey ← lastLetter (wi) + β ; Add the word to the list of words with the respective keys in the respective Hashtables. end; 290
  • 6. For each word in the word database, meter pattern is extracted first followed by letters at the rhyming spots namely, first, second and last positions (for monai, edhugai and iyaibu respectively). Keys for monai, edhugai and iyaybu are found out using the abovesaid equations. The word is added to the monai, edhugai and iyaibu hashtables with keys monaiKey, edhugaiKey and iyaybuKey respectively if those keys don’t appear previously in the tables. Else, the word is added to the list of words with that key. Hash-tables are chosen for the implementation as they have the retrieval complexity of O(1). 6. Results Figure 3. Word Indexed Vs Meter Rhyme Indexed Approach Indexing has brought about a drastic increase in the speed of retrieval of the words rhyming with a given word. Using the Word Indexed approach, the time complexity was O(α) where α was the total number of words. But after Meter Rhyme indexing, the complexity has become O(1) due to the use of hash table which is a very efficient data structure for retrieval. Rhyming words for a sample of 500 words were retrieved using both the approaches and the above mentioned graph was obtained. The Word indexed system tooK 875.47millisecond in an average while the Meter Indexed system tooK 1.90millisecond only.From the graph it can also be inferred that there are steep variations in the Word- Indexed approach while the Meter-Rhyme Indexed approach does not show such steep variations and is consistent. In terms of time efficiency, Meter-Rhyme indexed approach is evidently superior compared to Word-Indexed approach. In terms of space, it is not so efficient as each word will occur not once, but nine times in various Hash tables. 7. References Hisar Maruli Manurung: “An evolutionary algorithm approach to poetry generation”, Thesis for Doctor of Philosophy, University of Edinburgh, 2003. Sowmiya Dharmalingam., Madhan KarKy. “LaaLaLaa - A Tamil Lyric Analysis and Generation FrameworK” in World Classical Tamil Conference – June 2010, Coimbatore 291
  • 7. RamaKrishnan, A., S. Kuppan, and S.L. Devi. “Automatic Generation of Tamil Lyrics for Melodies” in NAACL HLT WorKshop on Computational Approaches to Linguistic Creativity. 2009. Colorado. K. Narayana Murthy "An Indexing Technique for Efficient Retrieval from Large Dictionaries", National Conference on Information Technology NCIT-97, 21-23 December 1997, Bhubaneswar. Eric Nichols, Dan Morris, Sumit Basu, Christopher Raphael, “Relationships between lyrics and melody in popular music”, ISMIR 2009, October 2009, Japan. Anna Babarczy, IldiKó Bencze, István FeKete, Eszter Simon, “The Automatic Identification of Conceptual Metaphors in Hungarian Texts: A Corpus-Based Analysis”, Proceedings of The seventh international conference on Language Resources and Evaluation (LREC), 2010, Malta. Xiao Hu, J. Stephen Downie, Andreas F. Ehmann, “Lyric Text Mining in Music Mood Classification”, ISMIR 2009, Japan. Jose P. G. Mahedero, Alvaro Martınez, Pedro Cano, “Natural Language Processing of Lyrics”, Proceedings of the 13th annual ACM internationalconference on Multimedia, New YorK, NY, USA, 2005. Rudolf Mayer, Robert Neumayer, Andreas Rauber, “Rhyme and style features for musical genre classification by lyrics”, Proceedings of the 9th International Conference on Music Information Retrieval (ISMIR’08), Philadelphia, PA, USA, September 14-18, 2008. Beth Logan, Andrew KositsKy, Pedro Moreno, “Semantic analysis of Song Lyrics”, IEEE International Conference on Multimedia and Expo (ICME), June 2004. Yael Netzer, David Gabay,Yoav Goldberg, Michael Elhadad, “GaiKu : Generating HaiKu with Word Associations Norms”, Workshop on Computational Approaches to Linguistic Creativity, CALC-2009 in conjunction with NAACL-HLT 2009, Boulder, Colorado. B. Logan, A. KositsKy, and P. Moreno, “Semantic Analysis of Song Lyrics”, in Proc IEEE ICME, 2004. 292