You need to train a natural language model to perform text classification on product descriptions that contain millions of examples and 100,000 unique words. You want to preprocess the words…

Question

Options

• ACreate a hot-encoding of words, and feed the encodings into your model.
• BIdentify word embeddings from a pre-trained model, and use the embeddings in your model.
• CSort the words by frequency of occurrence, and use the frequencies as the encodings in your
• DAssign a numerical value to each word from 1 to 100,000 and feed the values as inputs in your

• A
  3% (1)
• B
  85% (28)
• C
  3% (1)
• D
  9% (3)

Explanation

One-hot encoding (A) with 100,000 unique words produces extremely sparse, high-dimensional vectors that are computationally expensive and carry no semantic meaning. Assigning arbitrary numerical values (D) implies a false ordinal relationship between words. Sorting by frequency and using frequencies (C) encodes only commonality, not meaning. Pre-trained word embeddings (B), such as Word2Vec or GloVe, map each word to a dense, low-dimensional vector (e.g., 300 dimensions) where semantically similar words are geometrically close. These embeddings are especially well-suited for RNNs because they provide meaningful numerical input representations that the network can process sequentially, and leveraging pre-trained embeddings avoids training from scratch on the full vocabulary.

No community discussion yet for this question.