First, install the R client.
install.packages( "https://storage.googleapis.com/basilica-r-client/latest.tar.gz", repos=NULL )
Let's embed some sentences to make sure the client is working.
library('basilica') conn <- connect("SLOW_DEMO_KEY") sentences = c( "This is a sentence!", "This is a similar sentence!", "I don't think this sentence is very similar at all..." ) embeddings <- embed_sentences(sentences, conn=conn)
[[0.8556405305862427, ...], ...]
Let's also make sure these embeddings make sense, by checking that the correlation between the two similar sentences is larger:
print(cor(embeddings[1,], embeddings[2,])) print(cor(embeddings[1,], embeddings[3,]))
Great! You can see that the two very similar strings are much closer in this feature space. This same feature extraction can be use in a lot of different ways, including training regressions, classifiers, and clustering.