QB and semantic search (vs keyword based search)
Hello all
Happy New Years!
Even so i am not longer involved with QB support I am still interested of what QB can do. For instance in our company our service databases has 70K repairs cases (problems/solutions pairs) When we first started using QB back in 2006 we could not imagine having so many cases to search! But it is becoming a big issue. To the point where our engineers are less and less incline to search our QB to support their daily job which is to troubleshoot our customers instruments. The reason is even a targeted search returns 100's cases which need be look at manually to find what they looking for. Worst if they don't enter the correct term (misspelling) because they could miss the correct tool diagnoses that could have help them. That is why searching a large corpus of text (text fields...problem/solution) is not doable anymore. Our QB is gold mine but we need a better shovel :)
Anyway I have been looking at machine learning and NLP (for instance semantic search) to try improve our QB database search. Obviously the current "keyword" based search that QB is using is no longer working for us. I think using advanced NLP tool could help.
I am curious if anybody here has looked into semantic search? Or at least at more advance search methods. I think it is a big problem when you are dealing with free text inputs and have huge number of records. My hope of course is that QB will look into it and propose and integrated solution. Rather than us coming up with our solutions since this will means extracting our data and then use an external advance search process. It won't be seamless anymore I guess.
Will love to get the community ideas/though about this. Also, finding out if I am the only person who find the current search feature lacking when it comes to search for a large text corpus.
Thank you for your time.
Mo
------------------------------
Mohamed Bennouf
------------------------------
Happy New Years!
Even so i am not longer involved with QB support I am still interested of what QB can do. For instance in our company our service databases has 70K repairs cases (problems/solutions pairs) When we first started using QB back in 2006 we could not imagine having so many cases to search! But it is becoming a big issue. To the point where our engineers are less and less incline to search our QB to support their daily job which is to troubleshoot our customers instruments. The reason is even a targeted search returns 100's cases which need be look at manually to find what they looking for. Worst if they don't enter the correct term (misspelling) because they could miss the correct tool diagnoses that could have help them. That is why searching a large corpus of text (text fields...problem/solution) is not doable anymore. Our QB is gold mine but we need a better shovel :)
Anyway I have been looking at machine learning and NLP (for instance semantic search) to try improve our QB database search. Obviously the current "keyword" based search that QB is using is no longer working for us. I think using advanced NLP tool could help.
I am curious if anybody here has looked into semantic search? Or at least at more advance search methods. I think it is a big problem when you are dealing with free text inputs and have huge number of records. My hope of course is that QB will look into it and propose and integrated solution. Rather than us coming up with our solutions since this will means extracting our data and then use an external advance search process. It won't be seamless anymore I guess.
Will love to get the community ideas/though about this. Also, finding out if I am the only person who find the current search feature lacking when it comes to search for a large text corpus.
Thank you for your time.
Mo
------------------------------
Mohamed Bennouf
------------------------------
