Using a document as reference to boost results?

Hi,
I was wondering whether there is a way I can feed additional text (from a document) to get more accurate results?
For example, a user can upload a PDF that contains text that s/he thinks is relevant to the search. Then use something like TF-IDF to extract important words and feed them to something like a should query along with the actual query in a must query. Has anyone tried something like this? Has it worked? Any better way to achieve this?

Have you looked at the more like this query?

Yes @Christian_Dahlqvist But can I use it along with a user entered query? I believe this is the query you are referring to:

GET /_search
{
    "query": {
        "more_like_this" : {
            "fields" : ["name.first", "name.last"],
            "like" : [
            {
                "_index" : "marvel",
                "doc" : {
                    "name": {
                        "first": "Ben",
                        "last": "Grimm"
                    },
                    "_doc": "You got no idea what I'd... what I'd give to be invisible."
                  }
            },
            {
                "_index" : "marvel",
                "_id" : "2"
            }
            ],
            "min_term_freq" : 1,
            "max_query_terms" : 12
        }
    }
}

I understand the part of feeding in the artificial document, but I want to combine that with a user-entered query. Where do I put that?

@Christian_Dahlqvist It would be great if you can find some time to look into this. Thanks!

I am not sure it will get you where you want, but can’t you wrap this query and the user query in a Boolean query?

Something like the user query in a must clause and the extra keywords in a should clause?

Yes, something like that could potentially work.