Article Preview
TopBackground
In reference to the analysis by Delmonte et al. (2013), which describes and proposes analysis of poems based on rhythm and metrical structures wherein by the use of CMUdict (CMU, 2007) the words are converted to their phoneme formats and checked for same pronunciation. It evaluates the pattern of occurrence of the poem in terms of AABB, which are a set of predefined rules against which the system is evaluated. Kesarwani, (2017) proposes a similar method for organizing the poems into different classes based on rhymes, it takes into account the different types of rhymes that occur in the poem and rates the rhymes based on their type with a self-defined evaluation score metric. With the use of this score metric the difference between the gap of the poet can be drawn. The proposed system uses a similar detection method to evaluate the rhyme score for the poems, in accordance to the good reported accuracy of the same.
The style-based method for author and genre detection as previously described by Stamatatos et al., (2006) proposes to recognize the writing pattern in terms of disambiguation of the context and resolving the text into various chunks. They gather syntactic and token level information via the means of NLP tools and analyze the output to form features to categorize the text based on the authors. Rather using a parsing based approach, the proposed scheme follows a novel approach, which is based on analysis of simple human cognizable features.
Kaplan and Blei (2007) present an approach to visualize different poems as clusters and classifying them on the basis of quantitative analysis based on qualitative analysis which is similar to our thought process, yet instead of deriving the clusters, the authors focus on differentiating them on the basis of classification algorithms. Lou et al., (2015) also intend to classify the poems into various themes by the means of a classification algorithm, yet their feature collection method is based on the vocabulary of the poem, i.e term frequency and inverse frequency document measure for each of the poem.