room for improvement: https://onedx.find.sap.com

[JL23]

I certainly agree that searching with more terms is better than just using 1 word for a search but no matter how I search  I expect the same number of hits when I do this in the current site and in the beta site.

there are a few OSS notes where I remember the note number since really many threads are answered by referring to the OSS notes, so I tried a search by one of this numbers: 30656

The result with the current https://find.sap.com/ delivered 99 hits

the result with the new https://onedx.find.sap.com/ had only 8 hits, it was faster but that had probably to do with the fewer hits.

The current version shows full dates for the content found while the new search fails to do that.

I can guess that the date shown is the migration date to beta instead of the original posting date.

Still after sorting by newest first my blog from December 2014 was on top, which either means that the sort uses a different date than shown, or blogs are higher in the ranks than discussions.

It also looks as there are 2 english languages (English and En)

This is the same search via the current SCN search:

[agentry_src]

Hi JL,

How did the present SCN search results compare to the Google Custom Search ?  I have found numerous discrepancies there, especially with multiple word search terms. 

Cheers, Mike

SAP Technology RIG

[JL23]

I rarely (almost never) use the custom search.

I loved the older SCN search (some may remember, many still like the blog   and seem not to realize that his search has vanished) since I could filter by space.

I still find most what I  am looking for in the current SCN search, it is my first choice to search.

At least it gives me the option to ignore the wikis, I mean I can filter on more useful content like blogs, documents and discussions. Wikis is just a garbage place for me, I know about a handful excellent wikis but usually I do not find what I am looking for. To much empty pages with just a subject and links to a minimalistic other wiki pages. And the usual findings are just collections of transaction codes or copies from help.sap.com and SCN documents.

If I do not find anything relevant in the current SCN search then I use just Google. I have a small addon in the browser which allows me to mark a phrase and execute the search for this marked part via right mouse click in any of my preferred search places.

Often I search in SCN content where I or a certain person was involved, and for this I use http://scn.sap.com/search.jspa

[agentry_src]

Hi Juergen,

Interesting. I wish I could remember the older SDN search.  I don't think I used it enough to recall how well it worked.  Google was and pretty much still is my default search method.  I think I only got interested in the SCN Search when it was turned off during the initial SDN --> SCN transition. 

I periodically use SCN Search, usually after some posting indicating that huge improvements have been made.  I compare the result sets between the two and so far, always go back to the custom Google search.  Two main reasons are a more complete result set and better sorting inherent in the result set presentation. 

The Search in Open Beta has reached new lows in functionality and by itself should prevent a go live for the new 1DX environment.  When I get 2 Discussions (which really did not address the criteria very well) using the new search versus 491 mostly Discussions using the custom Google search with a the results ordered pretty much with the most appropriate content listed first, it makes it really obvious that there is a lot of work needed.

Cheers, Mike

[JL23]

It is hard to test and compare the results if most content is not migrated and can not be considered in the search results.

[agentry_src]

You may be right.  I did do a spot check (two) to see if the Discussions I found in the Google result set were in the Discussion archive and found both of them.  Not enough of a sample (as my statistician wife would tell me), but did nothing to dispel my doubts regarding the OB search functionality.   

[JasonLax]

The truth is, not all discussion are yet in the index. Since was posted in 2012, there probably have been a million discussion added to the community that create some challenges in performance and ensuring what is helpful gets to the top of the results.

I should have been more clear in my post and I've just gone back to updated now to indicate we are slowly adding the Discussions from the archive.  What's being prioritised for indexing are recent threads and the threads that are most heavily used: visited, linked to, appear in Google search, and other factors. (Yes, turns out the 80:20, if not 90:10 rule really applies to discussions.)

To be honest, I didn't think everyone would be comparing numbers like this because I was hoping topics experts could look the results for common queries and validate the results (i.e. answer the query or not). Eventually the beta search will return 491 results for the same query when the archive is fully indexed; however, who would go thru 491 results without changing the query to something more specific or applying filter?

Anyhow, the index will grow and perhaps I'll work on sharing a sample of top search queries to help guide the testing. 

[agentry_src]

Hi Jason,

Then the new search will be an improvement over the present SCN Search which also does not return the full number of items (by a wide margin) that the Google search returns.  I contacted the Search team, but apparently they have decided not to do any fixes until after the 1DX move is complete.  Not a good answer since this was quite a few months ago.

I will test again when you finish the move of Discussions.  Please let me know when the system is ready for a valid comparison.

Thanks, Mike

SAP Technology RIG

[JasonLax]

The full community archive is now being indexed.  The amount of archived discussion in search is over double what it was last week (500K to 1.3M) and ongoing operations will further bring the number 2M+ by mid-week.

[JasonLax]

Hi Juergen and thanks for the feedback.

• The 2 English languages appearing is know and being worked on (English and En).
• Dates also need improvement: The team is working with content owners to send over something more usable, such as created or last modified. When these are absent, the date the content was discovered by crawlers appears.
• The Beta Search still doesn't contain the full archive of discussion threads and this is the reason for fewer results. However, what has been added to Beta Search is more relevant with priority given to more recent and more used content: Quality over quantity.

So, aside from the UI and number of results, did the results provided answer the query? (Keep in mind that someone looking for a specific note will know to go to the note finder https://support.sap.com/notes)

[JasonLax]

Hi - Quick update on the dates.

The date information is now formatted on archive discussions and documents so you should begin seeing dates that match the last modified date in the results over the next 24-48 hrs.

Jason

[JL23]

>>>did the results provided answer the query?<<<

tough question...at least the result was not irrelevant to query.. give me some months till I really have a question and need to search

[JasonLax]

Hi - Dates are being captured now on discussions. I searched for "epm add-in performance" for example and saw variations in dates on Discussions.  It will take a few more days to recrawl things.

[JasonLax]

Hi - I've just activated the dates modified filter now.  Most content has a usable date, except mostly "Pages" (wiki and webpages from SAP.com).