Closed nchepanov closed 11 years ago
@0xd34df00d please take a look!
Already on it.
BTW this stuff required me to adjust MAX_BENI_LENGTH constant on barzer_barz.cpp:572. Is the version on the server the same as the master?
(and why the hell the local const is declared through an enum?)
the server version is master as of last sarurday
Sent from my iPhone
On Jun 5, 2013, at 3:48 PM, Georg Rudoy notifications@github.com wrote:
BTW this stuff required me to adjust MAX_BENI_LENGTH constant on barzer_barz.cpp:572. Is the version on the server the same as the master?
(and why the hell the local const is declared through an enum?)
— Reply to this email directly or view it on GitHubhttps://github.com/barzerman/barzer/issues/570#issuecomment-18970525 .
I'm not sure this can be fixed at all, source dataset is unnormalized.
please elaborate
Sent from my iPhone
On Jun 5, 2013, at 5:58 PM, Georg Rudoy notifications@github.com wrote:
I'm not sure this can be fixed at all, source dataset is unnormalized.
— Reply to this email directly or view it on GitHubhttps://github.com/barzerman/barzer/issues/570#issuecomment-18977068 .
i think we need to add boost for contiguous matches . lets discuss this ASAP
Implementing substring matcher sucks — we obviously want fuzzy matches, and fuzzy matching substring in another one is quite a PITA.
I suggest moving entirely to benisland stuff for this — we get the islands, their sizes and positions automatically and can boost our stuff accordingly. Implementing an efficient fuzzy matching algorithm is typically done via ngrams so it would lead to reinventing benisland.
lets start with strstr
Sent from my iPhone
On Jun 7, 2013, at 12:08 AM, Georg Rudoy notifications@github.com wrote:
Implementing substring matcher sucks — we obviously want fuzzy matches, and fuzzy matching substring in another one is quite a PITA.
I suggest moving entirely to benisland stuff for this — we get the islands, their sizes and positions automatically and can boost our stuff accordingly. Implementing an efficient fuzzy matching algorithm is typically done via ngrams so it would lead to reinventing benisland.
— Reply to this email directly or view it on GitHubhttps://github.com/barzerman/barzer/issues/570#issuecomment-19070684 .
Is this still an issue? Please triage.
spring cleaning
on this query beni works worse than sphinx search:
Варочная поверхность Nardi LG 430 AV N
BENI RESULT:
why the only right answer
Поверхность NARDI LG430AVN
has lower cover thanПоверхность NARDI LG 430 AV A
orПоверхность NARDI LG 430 AV DB