Ie in a BigTable table, for example, the table with the news. The header String is the body of the article Text. There is a search form on the news where you can enter keywords.
Actually in the API database queries with %LIKE%, though of course it is logical based on the ideology that each request needs to match your index.
Is API full text search — it takes a collection of documents and searches for words in text fields, substrings are not supported. Of course, you can generate different variants of the word, cutting off the prefix and end, but in fact it is also not the case. And even if you look at the full occurrence of the question — that I need to get the first 20 news for some keywords, I naturally can what is the number to pass the search API, but how many entities removed from the database? 20, 40? Again, unclear.
In General, what advise, how to do a search on the set keywords?
PS all this applies to the Java API, but in fact it does not matter much