Can the machine learning section of kibana help me to find words having the same meaning

can the machine learning section of kibana help me to find words having the same meaning (for example i would like to find the word good and also better, nice ..) how to do that (i ma using fscrawler and tesseract) thank you

Elasticsearch can handle synonyms without machine learning.

1 Like

does it support synonym for all languages ? and is it possible to use this functionallity in kibana web interface

The synonym capabilities make it possible to use synonym dictionaries in the text analysis process, but it doesn't provide these resources for any language. Synonyms are largely domain specific, so there is no one-size-fits-all approach. Depending on your needs you might be able to find good resources on the web (e.g. WordNet might be a good place to start your research, There are a number of ML and NLP techniques to find semantically correlated words. In a former project my team used methods described in Rapp: " The computation of word associations: comparing syntagmatic and paradigmatic approaches" with some success, but there is nothing ready-to-use in Elasticsearch ML as far as I know.

no i dont mean synonym like in a dictionnary but words having the samesd sens like when you say good, better , bad ....

how i cn do that (technically in elastic search

This is where the problem starts. "good", "better", "best" are comparatives and superlatives, the pair "good", "bad" is an an example of antonyms. Both are not what would pass as "synonyms" linguistically, although you can treat them technically as replace-one-by-the-other with what synonym filter do in Elasticsearch. But I'd be really careful about trying to do anything like this automatically if you're not entirely clear about what you want and need. See e.g. this great blog post about the dangers of using synonyms blindly.

1 Like

no, in fact i want to search a word like good and be able to see all the pdf that contains good or its synonyms this is my need

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.