I have a textual data set, containing product description. I need to classify these descriptions into different products. I am using TFIDF vectorization to vectorize the tokens into features for training. For any target label, is there a way to prioritize or increase or decrease the importance of certain features or words?

From the product description, I have identified certain keywords and vectorizing them to train the ML model. In certain descriptions, there could be multiple keywords that are relevant to multiple target labels. In this case, if I could give certain keywords, higher importance over another, then it will help the model to identify the right label or product. Is there a way to assign weights to these features?