Re: Text Similarity - testing on single words
23 views
Skip to first unread message
Message has been deleted
Roberto Santoro
Jan 31, 2020, 12:11:31 PM1/31/20
to Dandelion Support Forum
Hi Roy,
the Text Similarity API takes into consideration the entities extracted from the texts.
Generally speaking, the smaller the text, the more difficult it is to extract entities from it. Nonetheless, Dandelion API manages to extract entities even from small texts (and in this regard, the more specific the texts, the better): that's why it is able to recognize the similarity between "Labrador Retriever" and "Dobermann".
However, it is not intended to be used on single words, but on texts that, even if small, have a context. You should provide at least one or a few sentences.
Is your use case based on the similarity between single words? If this is not the case and instead you were just trying the API, I suggest that you try the Text Similarity API on sentences.
Best,
Roberto Santoro
Dandelion API team
the Text Similarity API takes into consideration the entities extracted from the texts.
Generally speaking, the smaller the text, the more difficult it is to extract entities from it. Nonetheless, Dandelion API manages to extract entities even from small texts (and in this regard, the more specific the texts, the better): that's why it is able to recognize the similarity between "Labrador Retriever" and "Dobermann".
However, it is not intended to be used on single words, but on texts that, even if small, have a context. You should provide at least one or a few sentences.
Is your use case based on the similarity between single words? If this is not the case and instead you were just trying the API, I suggest that you try the Text Similarity API on sentences.
Best,
Roberto Santoro
Dandelion API team
Reply all
Reply to author
Forward
0 new messages