One of the fundamental tasks of natural language processing (NLP) is to determine whether two words have a similar meaning to one another, and how closely they are related.
Why would you need this? If you’re analyzing threat reports, you’d want to be alerted to the word “bomb,” but wouldn’t you also want your text analytics solution to notify you about words like “explosive” and “incendiary device?”
As humans, we can recognize and compare words, but machines have difficulty decoding their meaning. That’s why NLP tools need word embeddings, which represent the meaning of words as numeric vectors that can be added, subtracted, and compared mathematically.
In this webinar, Babel Street Chief Scientist Kfir Bar explains the concepts around word embeddings and how they apply to real-life situations.
- A brief history of the advancements in semantic technology
- What do word embeddings enable us to do that we couldn’t before?
- How are word meanings calculated and compared?
- How word embeddings enable multilingual semantic search
- How semantic similarity boosts AI for extracting entities, matching names, and understanding events