MLS-C01 Question 98
Select 3You are tasked with building a machine learning model that predicts customer sentiment from product reviews. The reviews are written in English and include both text and emoji-based expressions. Which of the following approaches would help you effectively identify and extract relevant features from this dataset?
- A
Use a natural language processing (NLP) library to tokenize the text and convert it into word embeddings such as Word2Vec or GloVe.
- B
Apply image recognition techniques like Amazon Rekognition to analyze and extract features from emojis.
- C
Use one-hot encoding to represent emojis as categorical features in the dataset.
- D
Perform text preprocessing steps such as removing stop words, punctuation, and lowercasing text to clean the data.
- E
Leverage Amazon Comprehend to extract sentiment scores directly from the text reviews.
Show answer and explanation
Correct answers: A, D, E
Explanation
To effectively identify and extract features from text-based datasets, you need to leverage appropriate preprocessing techniques (e.g., tokenization, word embeddings, stop word removal) and AWS services like Amazon Comprehend for sentiment analysis. These techniques allow you to extract semantically meaningful features from both the text and emojis in the reviews. Image recognition tools like Amazon Rekognition and one-hot encoding are not suitable for text-based features like emojis, making them incorrect choices in this scenario.
- A. Correct.
Using an NLP library to tokenize the text and convert it into word embeddings like Word2Vec or GloVe is a common and effective way to represent textual data for machine learning models. These embeddings capture semantic meaning and are highly relevant for this task.
- B. Incorrect.
While emojis could carry sentiment information, they are not images. Image recognition tools like Amazon Rekognition are not suitable for analyzing emojis because emojis are text-based characters, not visual data.
- C. Incorrect.
One-hot encoding could represent emojis as categorical features, but it does not capture the semantic meaning of emojis in the context of sentiment analysis. Thus, this approach is not as effective for the given use case.
- D. Correct.
Text preprocessing steps help clean the data and improve the quality of features for machine learning. Removing stop words, punctuation, and lowercasing the text are standard practices that enhance NLP-based feature extraction.
- E. Correct.
Amazon Comprehend can be used to directly extract sentiment scores from the text reviews, providing you with a valuable feature for your machine learning model without requiring manual feature engineering.