September 4, 2017 at 2:56 am
Hi,
I am trying to implement LSA using sqlserver's fulltext and semantic search techniques. I have a table(Doc) comprising of 2000+ documents and I am doing a SEMANTICSIMILARITYTABLE on the table but it returns me only 10 rows, while the table has around 1000+ similar documents. Is this a know issue with SQL Server. I am using version 2016 and 2014. What are the way around it. Pls. help me, I am in the middle of some critical production release and this is not working.
Thanks
Surojit
September 5, 2017 at 7:38 am
surojitbhatt - Monday, September 4, 2017 2:56 AMHi,
I am trying to implement LSA using sqlserver's fulltext and semantic search techniques. I have a table(Doc) comprising of 2000+ documents and I am doing a SEMANTICSIMILARITYTABLE on the table but it returns me only 10 rows, while the table has around 1000+ similar documents. Is this a know issue with SQL Server. I am using version 2016 and 2014. What are the way around it. Pls. help me, I am in the middle of some critical production release and this is not working.Thanks
Surojit
Are you absolutely sure that your data does indeed meet SQL Server's definition of "similar", as opposed to your assumed definition thereof ? This may be one of those cases where you need to check your assumptions. Also, as we don't have any of your data to work with, we have no way to even test your scenario. If I were in your shoes, I'd start by looking at the actual documents and then at SQL Server's definition of "similar", and see if you can find two documents that actually meet the SQL Server definition, and see if SQL Server will agree with you. If you're going to rely on that kind of a feature, it's usually a good idea to test that you can indeed get the result you're looking for back in the planning phase of the project.
Steve (aka sgmunson) 🙂 🙂 🙂
Rent Servers for Income (picks and shovels strategy)
September 6, 2017 at 5:13 am
Hi,
I am sure that their exists more then 100 mathcing documents as I have knowingly populated the same document in other records and its more then 500+.
I also hit a discussion thread in ms https://docs.microsoft.com/en-us/sql/relational-databases/system-functions/semanticsimilaritytable-transact-sql which talks of a similar issue.
I am evaluating various tools and I thought SQL could help me attain this because of its inbuilt support and considerable first query time, enough for my application.
Looking forward for your suggestion.
Thanks & Regards
Surojit
September 6, 2017 at 2:21 pm
surojitbhatt - Wednesday, September 6, 2017 5:13 AMHi,I am sure that their exists more then 100 mathcing documents as I have knowingly populated the same document in other records and its more then 500+.
I also hit a discussion thread in ms https://docs.microsoft.com/en-us/sql/relational-databases/system-functions/semanticsimilaritytable-transact-sql which talks of a similar issue.
I am evaluating various tools and I thought SQL could help me attain this because of its inbuilt support and considerable first query time, enough for my application.
Looking forward for your suggestion.Thanks & Regards
Surojit
If you read the user's question, it looks like another MS "feature" that's not documented. Not sure if anyone has tried to post this as a CONNECT item. You might try that route, but in the meantime, it looks like 10 is an internal limit. Once again though, I go back to properly researching the feature to see if anyone has had any trouble with it before you rely on it as a methodology, and especially when it's likely to be a rarely used functionality.
Steve (aka sgmunson) 🙂 🙂 🙂
Rent Servers for Income (picks and shovels strategy)
Viewing 4 posts - 1 through 4 (of 4 total)
You must be logged in to reply to this topic. Login to reply