'Combining seed words in word embedding [closed]
I have a word embedding model that I am using to flag reviews. Consider the seed word "animal". This model will generate the synonyms for the word "animal", then flag the reviews containing those synonyms. I would like to increase the number of synonyms, so that I may flag more reviews.
My understanding of Word2Vec implies that, for example, if the data contains no instances of "animal" and "fauna" within a certain number of words of each other, Word2Vec will not list "fauna" as a synonym of "animal".
Is it possible to combine seed words?
This would run the word embeddings model once on "animal", and again on "fauna", then find the synonyms of both. What is the best way to do this? Two ideas are
- union the synonyms of both
- average the two seed word vectors
Are either of these valid or is there another way to do this?
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|