a) Explain why latency is a critical issue for chatbots. (2 marks)
b) Describe two strategies that could be implemented to reduce the chatbot’s latency. (4 marks)
Latency affects user experience. Strategies include optimizing NLP pipelines and using powerful hardware.
a) Explain one major limitation of RNNs in chatbot performance. (3 marks)
b) Compare how LSTM and Transformer models address these limitations. (5 marks)
RNNs suffer from vanishing gradients. LSTMs improve memory, while Transformers enhance efficiency using self-attention.
a) Define three characteristics of a high-quality training dataset. (3 marks)
b) Explain how dataset bias can impact chatbot performance. (3 marks)
c) Describe two strategies to improve dataset quality and diversity. (4 marks)
High-quality datasets are domain-specific, diverse, and unbiased. Bias can lead to discrimination. Strategies include using synthetic data and audits.
a) Identify and explain two ethical concerns associated with chatbot deployment. (4 marks)
b) Suggest one method to address each concern identified. (2 marks)
Concerns: Data privacy and bias. Solutions: Encryption for privacy, bias testing for fairness.