Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Inproper search results #509

Open
peter-empen opened this issue Nov 12, 2017 · 9 comments

Comments

@peter-empen
Copy link

@peter-empen peter-empen commented Nov 12, 2017

This is related to #389 but I feel even worse since misleading:

  1. search for "graph": the second hit is "lewismj / cucumber". This project has nothing to do with graphs, right? Also, I'd expect "scala-graph" to pop up in the hit list but it doesn't. Note that "gremlin-scala" is a proper hit.
  2. search for "scala-": no project starting with "scala-" is found. It seems there is an issue with dashes.
  3. search for "scala-gr": "scala-arm", "scala-uri" are now contained, "scala-graph" is still not.
  4. to get "scala-graph", having some 360 stars, you need to fully spell it out.

Looking at "scala-graph" do you see any issue that causes scaladex search to bypass the project?
What is the reason for "gramlin-scala" being included while "scala-graph" not?
Does scaladex use any black list, if so is it related?
What action is to be taken to get "scala-graph" listed on a search for "graph"?

@MasseGuillaume

This comment has been minimized.

Copy link
Member

@MasseGuillaume MasseGuillaume commented Nov 16, 2017

  1. This is due to an artifact called graph: https://index.scala-lang.org/lewismj/cucumber/graph/0.0.6?target=_2.12. When you search for graph you see a: Did you mean topics=graph?.

  2. The search result comes from the query-string-syntax of elasticsearch. The query would be: https://index.scala-lang.org/search?q=repository%3Ascala-*

  3. Similar to 2: https://index.scala-lang.org/search?q=repository%3Ascala-gra*

I think https://index.scala-lang.org/search?q=&topics=graph gives the best result for this. Maybe the color contrast in not high enought for the warning message: Did you mean topics=graph?.

I agree that all of this is suboptimal. Search relevance is a hard problem and we are still willing to improve on this.

@eed3si9n

This comment has been minimized.

Copy link

@eed3si9n eed3si9n commented Dec 15, 2017

Clicking on Akka link from the front page, I get the following..

search__

akka/akka doesn't show up until page 4. Could we please use GitHub star as the default sort ordering for now?

@tpolecat

This comment has been minimized.

Copy link

@tpolecat tpolecat commented Dec 15, 2017

Could we please use GitHub star as the default sort ordering for now?

👍 from me. The "relevance" ordering still seems basically random to me. The database topic lists Quill and doobie dead last.

@peter-empen

This comment has been minimized.

Copy link
Author

@peter-empen peter-empen commented Jan 13, 2018

Looking at this again, there is an improvement concerning 1. Did you change anything?
You mentiond a message"did you mean topic..." but I have never seen such a message in Firefox.

@xingyif

This comment has been minimized.

Copy link

@xingyif xingyif commented Jul 1, 2018

@smarter @eed3si9n what's the decision on this? are we using Stars as the default sorting result?

@eed3si9n

This comment has been minimized.

Copy link

@eed3si9n eed3si9n commented Jul 1, 2018

If it's up to me I'd sort by GitHub stars, augment it with "featured" manual curation. But you'd have to ask someone from Scala Center, not me.

@MasseGuillaume

This comment has been minimized.

Copy link
Member

@MasseGuillaume MasseGuillaume commented Jul 1, 2018

Sorting with starts is not the way to go. It will give even more unrelated results. It will simply put projects with more stars on the top. For example, if you search for json, you will find yourself with Spark.

@eed3si9n

This comment has been minimized.

Copy link

@eed3si9n eed3si9n commented Jul 1, 2018

Using topics=json as suggested by the front page, and also by the "did you mean?" feature, it shows scalatra/scalatra on top using Relevant. Curiously json4s or circe is nowhere to be found in the first few pages.

Using Stars, scalatra/scalatra is still on top, but at least you have circe, spray-json, jawn, play-json, scalajson, and json4s on page 1. It's actually pretty usable as the sorting. It would certainly be better if you can manually bump down web frameworks like Scalatra and Finch, but Stars is doing a better job at finding useful things.

json

@MasseGuillaume

This comment has been minimized.

Copy link
Member

@MasseGuillaume MasseGuillaume commented Aug 9, 2018

Let's keep one ticket for relevance: #389

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
5 participants
You can’t perform that action at this time.