Query Recall Vs Precision

[RabBit_BR (andre.coelho)]

Hello everybody.

I would like some help to maintain the accuracy of my query and still have a minimal recall of results.

Today I have some documents that have generic terms, and these terms retrieve products that
outside the search context.

For example:

Query:

"should": [
            {
              "match_phrase": {
                "title.raw": {
                  "query": "#$query##",
                  "slop": 2,
                  "boost": 100
                }
              }
            },
            {
              "match": {
                "title": {
                  "query": "#$query##",
                  "operator": "AND",
                  "boost": 100
                }
              }
            },
            {
              "multi_match": {
                "query": "#$query##",
                "fields": [
                  "description.pt-br^1.0",
                  "title.pt-br^1.0"
                ],
                "type": "best_fields"
              }
            }
          ],

Documents:

Course Python
Course Java
Course Photoshop
Course Advanced Java
java for beginners
Java Ninja
Course ....
.
.
More docs name with Course

When I run my query to search for docs related to the term "course java" I have these
results:
Course Java
Course Advanced Java
java for beginners
Java Ninja
Course Python
Course Photoshop
.... more doc with "Course"

Note that it retrieves Python Course and Photoshop Course due to the query construction itself that uses multimatch.

Is it possible to limit the recall of documents with the term "course"?