Which statement about embeddings is true?

Which statement about embeddings is true?

• A. They only capture frequency data
• B. They replace the need for any form of syntax
• C. They capture semantic and syntactic relationships between words
• D. They are used only for sentiment analysis

Answer: (C)

Word embeddings place words in a dense vector space where proximity reflects how words are used and what they mean. Learned from large text, these vectors capture both semantic relationships (meaning and related concepts) and syntactic patterns (how words function grammatically). That combination lets the model recognize that related words are close together and that certain grammatical relationships show up as consistent directions in the space. A classic illustration is that king minus man plus woman yields a vector close to queen, demonstrating how semantic content and syntactic form can be represented algebraically in the same space. These representations come from context patterns, not just raw frequency counts, and they support a wide range of tasks beyond sentiment analysis, such as translation, information retrieval, and clustering.

Word embeddings place words in a dense vector space where proximity reflects how words are used and what they mean. Learned from large text, these vectors capture both semantic relationships (meaning and related concepts) and syntactic patterns (how words function grammatically). That combination lets the model recognize that related words are close together and that certain grammatical relationships show up as consistent directions in the space. A classic illustration is that king minus man plus woman yields a vector close to queen, demonstrating how semantic content and syntactic form can be represented algebraically in the same space. These representations come from context patterns, not just raw frequency counts, and they support a wide range of tasks beyond sentiment analysis, such as translation, information retrieval, and clustering.