Language Model and Word Embeddings Quiz

What is the standard way to represent meaning in NLP?

• Using tf-idf

• Using dense vectors

• Using sparse vectors

• Using word vectors

Why are word embeddings used in natural language processing?

• To work with high-dimensional datasets

• To capture synonymy and similarity between words

• To provide arbitrary encodings for words

• To represent words as discrete symbols

What do vector space models (VSMs) do?

• Represent words as discrete symbols

• Create arbitrary encodings for words

• Map semantically similar words to nearby points

• Work with low-dimensional datasets

What is the main advantage of dense vectors over sparse vectors?

• They are more common in NLP tasks

• They are easier to use as features in machine learning

• They require less memory storage

• They capture synonymy better

Which method is known for being very fast to train and has code available on the web?

• Word2vec

• Glove

• Fasttext

• Tf-idf

What is the main idea behind Word2Vec?

• Counting the occurrences of each word

• Predicting rather than counting

• Using hand-labeled supervision

• Using running text as explicitly supervised training data

How is the similarity between words modeled in Word2Vec?

• Using cosine similarity

• Using dot product

• Using unigram frequency

• Using logistic regression

What is the main purpose of using negative samples in the skip-gram algorithm?

• To improve the classifier performance

• To decrease the training time

• To create a balanced dataset

• To increase the training time

How are word embeddings evaluated?

• By comparing to human scores on word similarity tasks

• By measuring the length of the vectors

• By using unigram frequency

• By counting the occurrences of each word

What is the main advantage of using Tensorflow for Word2Vec?

• It provides a variety of options for word embeddings

• It allows for easy installation and usage

• It is the most basic classifier used in machine learning

• It is the fastest method for training word embeddings