Poetry is a special cultural heritage that has been around for more than 2,000 years. Its popularity manifests itself in many areas of daily living. For example, artists use poems as a means of expressing emotion, political views, or self-expression. Unlike free language, poems are designed to be elegant, concise, and aesthetic.
Composing poems is a challenging task. This is because you need to follow a set of phonological, semantic, and structural requirements, so only some scholars can master the art to organize or manipulate terms.
The Foundation of Automatic Poem Generation
With the advancements in machine learning and artificial intelligence, computers can now help humans to create poems. They can generate appropriate word combinations from a text corpus.
Programs can now learn, remember, and recognize rules or patterns given by an electronically stored and structured set of texts. Computational intelligence serves as the motivation behind automatic poem generation.
Understanding the Issues in Making Poems
Poetry generation involves regular phonetic and syntactic patterns where metrics, rhyme, rhythm, and other features such as figurative or alliteration language plays a vital role. When it comes to writing poems manually, the poet needs to break some grammar rules that are often used in producing natural language text. The following are the issues that need to be considered:
- The occurrence of correlated linguistic phenomena that requires the consideration of lexis, semantics, and syntax.
- The absence of a well-defined tone and message.
- The need for rich resources to satisfy syntax, semantics, and phonetics
Writing poetic text is a difficult task. You must follow many rules, and at times, break some of them. But, with the use of machine learning and other IT paradigms, creating poems is now a breeze.
The Different Poem Generation Systems
This article presents to you the history, approaches, techniques, categorizations, and goals of automatic poetry generation. It is possible to group the automatic poem generation systems based on the techniques and approaches used:
- Template-based poetry generation
With this approach, poetry templates are filled with words from a dictionary to fill a set of rhythmic and syntactic constraints. The following are the most notable template-based poetry generation:
- Poetry Creator
- ALAMO group
- Generate and test approaches
For generate and test approaches, random word patterns and sequences are generated based on defined requirements that involve other semantic and formal constraints. The WASP system, Tra-La-Lyrics, and Manurung’s chart system follow this approach.
- Cased-based reasoning approaches
Poems that are available on the internet are retrieved. The content being taken depends on the set requirements of the user. The retrieve texts are then used to fill the required content. COLIBRI and ASPERA are examples of this technique.
In this approach, the produced output is based on evolutionary computing. Evolutionary computation follows the concepts of genetic inheritance and natural selection. It is suitable for poetry generation, like how authors write poems. Manurung’s McGonagall and POEVOLVE are two systems that are based on the evolutionary approach.
Poems, whether manually or automatically created, must incorporate all the three properties mentioned below:
The poem must convey a message that is meaningful under certain forms of interpretation.
It must obey linguistic conventions established by a given lexicon and grammar.
The poem must manifest poetic features.
Except for the properties, automatic poem generation systems are also classified based on the goals that they are designed to achieve.
This is the system that connects randomly generated words together. Word salad doesn’t follow any grammatical rules because there are no properties embodied in the generated poem.
- Template and grammar-based
Words are selected and taken from a certain lexicon to fill the gaps in poem templates.
The chosen words are generated based on a pre-defined form, such as the sonnet or the haiku. Form-aware follows metrical rules to accomplish the goals given to it.
The generated poetic texts exhibit all the three properties of poeticness, grammaticality, and meaningfulness.
The automatic generation of poems is an interesting and complex technology, for it involves different categorizations and approaches, and several levels of languages. While some of the techniques above rely on pre-defined metrics, others are concerned with semantics and syntax. If you’re interested in trying out an automatic poem generator, visit generateapoem.com. This website offers a free automatic poem service. It employs statistical approaches to generate, analyze, translate, and provide you a creative composition that is based on your specified poem format.