用skip-gram或者CBOW训练自己的词向量

This GitHub repository contains implementations of three popular word embedding techniques: Singular Value Decomposition (SVD), Continuous Bag of Words (CBOW), and Embeddings from Language Models (ELMO). Word embeddings are a fundamental component of natural language processing and are essential for various text-based machine learning tasks.

This project implements Word Embedding using the Continuous Bag of Words (CBOW) method for natural language processing tasks. The program processes PDF files, tokenizes text, trains a Word2Vec model using CBOW, and evaluates the cosine similarity between selected word pairs from the document.

I performed sentiment analysis aimed at determining the sentiment of 50000 imDB movie reviews, whether they are positive, negative, or neutral. I employed various NLP approaches including lexicon based approaches, machine learning models, PLM models, and hybrid models, and assessed the performance on each type of model.

Code for implementation of word embeddings from scratch in python using Frequency-based Embedding(Co-occurrence Matrix method) and Prediction-based Embedding method(Word2vec method)

A Basic Word2Vector WordEmbeddings Model. With image2Vector and Audio2Vector Encoding and decoding. (Audio is not great but works NEeds improvement - but can be reconstructred)

The Use Of Classical Classification to Distinguish between 16 MBTI given a vectorized text using CBOW, BERT Models vs Classification using The LSTM model