AI-Powered Poetry Generator using TikTok User Data

Inspiration

The idea for this project stemmed from my fascination with both artificial intelligence and creative expression through poetry. I was inspired by the potential of using AI to analyze and generate poetic verses based on real-world data from TikTok, a platform rich with diverse user-generated content.

Learning Experience

Throughout the development of this project, I gained valuable insights into natural language processing, machine learning models (specifically Markov chains and LSTM networks), and integrating external APIs like TikTok's API. I also deepened my understanding of syllable counting, rhyming patterns, and text generation techniques.

Project Development

I started by collecting and preprocessing textual data from TikTok, focusing on hashtags related to various themes. The data preprocessing involved cleaning and formatting the text for further analysis. Using Python, I utilized libraries such as TikTokApi, markovify, pronouncing, and keras to build a robust AI model capable of generating poetic lines.

Challenges Faced

One of the main challenges was optimizing the LSTM network to produce coherent and aesthetically pleasing poetry. Balancing syllable constraints and rhyme schemes while maintaining semantic coherence posed significant algorithmic challenges. Additionally, working with the TikTok API and ensuring data consistency and quality required careful handling.

Built With

• ai
• generative
• kaggle
• keras
• lstm
• markovify
• natural-language-processing
• network
• neural
• pronouncing
• python
• rnn
• tiktokapi