Tuning the Multipliers for Rhyming

I am still focusing on rhyming the final words in a haiku. I decided to change how the synonym accuracy was affecting the final value. Instead of adding p(e|f) and p(f|e), I decided to multiply them together so that accuracy played a larger role in the final value.

To compare differences between the original algorithm and the altered algorithm, I compared the results from the haiku used in my last poetry post: Rhyming Poetry with Haikus. Here are the differences, where A1 stands for Algorithm 1 (or the original algorithm) and A2 stands for Algorithm 2:

Original Poem:                 Translated Poem (A1):      Translated Poem (A2):

at the age old pond          at the age old ocean          at the age old pond
a frog leaps into water    a frog leaps into urine      a frog leaps into water
a deep resonance             a deep reception                a deep resonance

A1 – {‘ocean urine reception’: 51.851460000000003}

A2 – {‘pond water resonance’: 45.0}

I also tested the differences using a different haiku. The results follow:

Original Poem:                    Translated Poem (A1):        Translated Poem (A2):

Falling to the ground          Falling to the sun                Falling to the ground
I watch a leaf settle down  I watch a leaf settle gun     I watch a leaf settle down
In a bed of brown               In a bed of brun                  In a bed of brown

A1 – {‘sun gun brun’: 86.198834976800001}

A2 – {ground down brown’: 39.0}

The above translation had a value of approximately 86.19, which is significantly lower than the translation I believe fit the best – {‘ground down brown’: 108.0}. In my opinion, the original poem already had a good rhyming scheme and therefore did not need to be changed. The original algorithm focuses too much on rhyming and not enough on meaning preservation.

Multiplying the synonym accuracy together results in a translation that now focuses too much on meaning preservation and not enough on rhyming. I need to find a balance between the two.

I played around with the multipliers and it seems the translations now have a good balance between meaning preservation and rhyming. I multiplied the translation accuracy and the rhyming multipliers. I also altered the rhyming scheme multipliers:

1: both phonemes of both words are the same

2: one word only had one phoneme but the phoneme rhymed with the last phoneme of the other word

1: only one of the phonemes of each word are the same

10: no similarity in phoneme pairs

I assigned a value of ‘1’ to the approximate rhymes because professional poets use approximate rhymes more than amateur poets. There is a further explanation of the differences between professional and amateur poets in an article I read that: Ranking Contemporary American Poets.

The rhyming haiku results were much better than my original algorithms. For example, the first poem was translated as:

at the age old dugout
a frog leaps into fleet
a deep resonant

{‘dugout fleet resonant’: 78.9253794408}

‘Dugout’ is a depression in the ground, ‘fleet’ is another word for a creek or stream, and ‘resonant’ is just an alternative of the word ‘resonance.’ This translation proves much more accurate than the one from the original algorithm and the approximate rhyming is better than the original poem.

The second poem was translated as:

Falling to the reason
I watch a leaf settle down
In a bed of brown

{‘reason down brown’, 25.4001052118}

‘Reason’ may not be a truly accurate translation for ‘ground’ but this algorithm results in a nice poem with good approximate rhymes.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: