I am a newbie with elasticsearch and NEST, and in my exploration phase. Currently I am ingesting some attachment types to search and highlight from its content field (analyzed text). Now I want to implement autocomplete suggestions so that if user types some phrase it suggests the top N best-suited completion matches for that phrase from the attachment contents. I looked into some edge ngrams examples but they are dealing with fields like title, name etc. In my case these could be large text fields from which the phrase has to be searched and autocompleted.
Any working example would be great, if not at least if some one can guide me towards the possible approaches.
I am using the ingest-attachment plugin that extracts the text out of the attachments and populates the index. Also used some analyzers and mapped them. The search/suggest etc. works fine and will continue to make them more precise. But the part I am stuck with is completion for partially typed phrases. I agree that the phrases should best be domain specific. But to start with, what I am looking for is a way to scan the extracted texts and keep the most important keywords (ranked based on some scoring strategy) ready. I think significant text aggregation does exactly the same: https://www.elastic.co/guide/en/elasticsearch/reference/master/search-aggregations-bucket-significanttext-aggregation.html. But unfortunately it is not available with NEST(22.214.171.124). Will it be available in any future release or a CI build?