Exploring-Word-Embeddings-with-Word2Vec

This project implements two key NLP tasks using the pre-trained word2vec-google-news-300 model via the Gensim library:

Tasks

1. Replace with Similar Words

Given a sentence and a list of word indices, this task retrieves the top 5 most similar words for each indexed word and replaces each with one randomly selected alternative.

• Function: replace_with_similar(sentence: str, indices: List[int]) -> Tuple[str, Dict[str, List[Tuple[str, float]]]]
• Returns: A new sentence and a dictionary of similar words with similarity scores.

2. Find Similar Sentences

Computes sentence embeddings by averaging word vectors and finds the most similar sentences to a query based on cosine similarity.

• Functions:
  • sentence_vector(sentence: str) -> Tuple[Dict[str, np.ndarray], np.ndarray]
  • most_similar_sentences(file_path: str, query: str) -> List[Tuple[str, float]]

Dependencies

• Python 3.x
• gensim
• numpy

Pre-trained Model

import gensim.downloader
model = gensim.downloader.load("word2vec-google-news-300")