Revisit unfielded query mapping to author search

[aaccomazzi]

Currently a query of "Kepler 1362 b" gets rewritten by our modified edismax into something that includes author:b, *. This is clearly undesirable, so we should reconsider the rewriting rules to avoid this chain of events

[JCRPaquin]

Do we want strings in quotes to not be expanded at all? That is, we'd broadcast them as-given across all fields that an unfielded search would match. Even then I think the author name tokenizer would likely cause the same issue?

[aaccomazzi]

I'm not sure I understand the question, so let me restate it as such and see if we agree. An quoted unfielded search ("A B C") IMHO should generate a edismax query across fields where the quotes are preserved for each one (something like author:"A B C" | title:"A B C" | abstract:"A B C" ...). So if I understand the question correctly, I'm saying that they should be broadcast as phrases across fields (but I'm still not sure what you mean by "expanded" above).

The problem of this query really lies in the fact that the above query generates, among others, a search for author:"Kepler 132 b" which then is further broken down into something that includes author:"b, *". The first step we should take is to review the expansion of the author queries at least for names that are not preformatted as "Last, First"