Refined Global Word Embeddings Based on Sentiment Concept for Sentiment Analysis

Sentiment analysis is a significant area of study in natural language processing that finds extensive use in journalism, politics, and other domains. In sentiment analysis, word embeddings are important. The sentiment lexicons are directly incorporated into conventional word representation using the current senstiment embeddings techniques. This sentiment representation technique is unable to offer precise sentiment information for words in many situations since it can only distinguish the sentiment information of distinct words, not the same word in several settings. To address the issue, this research suggests the sentiment idea. First, based on the word context, we determined the best sentiment idea for words in the Microsoft idea Graph. Then, using the multi-semantics sentiment intensity lexicon that we built in this research to ensure accurate. Embedding of sentiment information and provide more accurate semantics and sentiment representation for words, we extracted the sentiment information of words under the optimal sentiment concept. Ultimately, we integrated enhanced word embedding techniques to produce a more thorough word representation. The validity of the word embeddings method based on the sentiment notion suggested in this work is confirmed by comparing it with traditional and sentiment embeddings methods on six typical datasets.