Although the word2vec concept is very efficient, there are some points that you may find complex or challenging. Here, I will propose the most common challenges. Those points are listed as follows:
- The word2vec model is easy to develop, but difficult to debug, so debug ability is one of the major challenges when you are developing a word2vec model for your dataset.
- It does not handle ambiguities. So, if a word has multiple meanings, and in the real world we can find many of these kinds of words, then in that case, embedding will reflect the average of these senses in vector space.
