My database is currently storing information related to user searches. For example we have a User and Term vertex. User has a property called employeeId while Term has a property called "q" which will contain the search query. We draw an edge called 'searched' between them. What I would like to be able to do is suggest other searches to the user which might be related. I have a query which I feel sort of does the trick by doing a traverse from a term and ordering by depth and count. I'm sure there might be better way out there to do the query, or find better results.
User --searched--> Term
In this snippet below I would send in a term that got searched for, and use that in a like search to find terms which also contain that word.
@orient_query = "SELECT $depth, q, in().size() AS count FROM (TRAVERSE * FROM (select from Term where q.toLowerCase() LIKE 'aluminum') STRATEGY BREADTH_FIRST) WHERE @class = 'Term' AND $depth <> 0 ORDER BY $depth ASC, count DESC"
I guess my main concern is that there is a much better way out there to do this and find better results. My original thought process was that we needed a combination of depth in the traversal vs counts of how many times that search came up.
Edit: What I'm also seeing that other searches which contain the word "aluminum" like "aluminum alloy" are showing up really far down the results since their depth is higher.
