Understanding the intent of the customer is key to implementing a successful chat-bot experience for end-user. Here, we have are in-charge of the task of Customer Intent Classification.
As part of this assignment, you are provided with the ATIS (Airline Travel Information System) dataset with two CSV files: atis_intents_train.csv and atis_intents_test.csv.
- Fine-tune a pre-trained Transformer model of your choice using the atis_intents_train.csv and evaluate your performance on the atis_intents_test.csv. Please feel free to use any Fine-Tuning approach to solve this problem.
- In the real-world, getting the labels for the entire dataset can be quite expensive. In such cases, semi supervised/few-shot text classification research papers and models can be your ally! For this problem, please do your own Literature Review and choose a suitable research paper to implement using just 10% of the data in the atis_intents_train.csv and evaluate your model’s performance on the atis_intents_test.csv.
Guidelines:
- I have used TensorFlow/Keras
- For evaluation, Accuracy, F1 Score and F1 Score
Algorithm used: LSTM:
Research Paper: https://arxiv.org/abs/2001.09330 https://www.sciencedirect.com/science/article/pii/S1877050918320374