Podcast
Questions and Answers
What is LSA and how does it approach word representation?
What is LSA and how does it approach word representation?
LSA stands for Latent Semantic Analysis. It factorizes the term-document-matrix to obtain lower-dimensional dense features as word representations.
How can we obtain word representations through document occurrences?
How can we obtain word representations through document occurrences?
One approach is to use the term-document-matrix to represent words based on their occurrences in different documents.
Explain the concept of neighboring words in obtaining word representations.
Explain the concept of neighboring words in obtaining word representations.
Neighboring words can be used to create cooccurrence vectors that represent the relationships between words based on their proximity in text.
What is an n-gram?
What is an n-gram?
Give an example of bigrams in a text.
Give an example of bigrams in a text.
How can n-grams be beneficial for other languages?
How can n-grams be beneficial for other languages?
What is the motivation behind using word embeddings?
What is the motivation behind using word embeddings?
Why do we want word representations with lower dimensionality than our vocabulary?
Why do we want word representations with lower dimensionality than our vocabulary?
How can we interpret document vectors in the bag of words model?
How can we interpret document vectors in the bag of words model?
What is the basic idea behind Word2Vec?
What is the basic idea behind Word2Vec?
What are the two main configurations for training Word2Vec?
What are the two main configurations for training Word2Vec?
How many dimensions are typically used in a single layer network for Word2Vec?
How many dimensions are typically used in a single layer network for Word2Vec?
What is the purpose of mapping every word to a layer in Word2Vec?
What is the purpose of mapping every word to a layer in Word2Vec?
Does Word2Vec map words or documents to a layer in the neural network?
Does Word2Vec map words or documents to a layer in the neural network?
