This repository contains the solutions to three problem statements completed during the hackathon. Each problem statement is categorized based on its difficulty level: Easy, Moderate, and Hard.
Problem Statement Description: The task was to develop an AI system capable of providing autocorrect suggestions for English words. The system should recognize misspelled words and suggest the correct spelling. It should handle a wide range of words and common spelling mistakes.
Solution: For this problem statement, the S2S (Sequence-to-Sequence) model was utilised. It is a form of neural network architecture that can be applied to fix or boost the precision of text that is automatically created. The model creates a corrected output sequence from a potentially incorrect input sequence.
Reference: The GeeksforGeeks tutorial on autocorrect features using NLP in Python was used as a basis for the implementation.
Problem Statement Description: The task involved developing an AI system capable of performing sentiment analysis on the posts and comments of any given LinkedIn page or post. The system should categorize the text as positive, negative, or neutral and provide an overall sentiment distribution.
Solution: To extract sentiment analysis features and categorise the sentiment as positive, negative, or neutral, Hugging Face's pre-trained model which was based on RoBERTa based on transformers was applied. This model was trained on 58 million Twitter tweets. The streamlit library was used to produce the graphical portion of the system and show the findings once the categorization of emotion was extracted. Due to the size of the dataset used for training, the model's prediction accuracy was roughly 99.5%.
Reference: The research paper on sentiment analysis was used as a reference to guide the implementation.
Problem Statement Description: The task was to train a ChatGPT model to function as a Gratitude Assistant. The model should generate thoughtful, compassionate, and personalized responses that express gratitude based on the user's input. The model should understand the context of the conversation and generate appropriate gratitude responses.
Solution: For this problem statement, the OpenAI ChatGPT 3.5 model was utilized. The model was trained to generate gratitude responses that are contextually accurate, convey genuine gratitude, and are personalized to the user's input.
The repository is organized as follows:
/
├── Easy-Autocorrect-Suggestion/ (Contains the solution for the Easy Problem Statement)
├── Moderate-Sentiment-Analysis/ (Contains the solution for the Moderate Problem Statement)
├── Hard-Gratitude-Assistant/ (Contains the solution for the Hard Problem Statement)
└── README.md (You are here!)
Each sub-folder contains the code, data, and any additional resources required for the respective problem statement.
To use the solutions provided in this repository, follow the instructions provided in each sub-folder's README.
We would like to express our gratitude to the hackathon organizers for providing us with these interesting problem statements. Additionally, we appreciate the references and resources that helped guide our implementations.
- Mohammed Valiya - @Mohammedvaraliya
- Subhashish Nabajja - @subhashishnabajja
- Awais Shaikh - @awais-shaikh
- Vaibhav Patel - @VaibhavProject-1
This project is licensed under the MIT License. Feel free to fork, modify, and use the code for your own purposes.
Happy hacking! 🚀