chang detect find distribut shift data stream databas explor databas dynam inform qualiti network monitor analysi databas explor databas dynam mine network log inform qualiti challeng stewardship stage inform enterpris mine exploratori data mine data clean data qualiti knowledg engin data qualiti data clean overview databas explor bellman larg industri scale databas tend poorli structur dirti confus reason disord applic domain poorli structur dirti confus data analyst call mine clean analyz databas articl describ type problem encount tool techniqu develop address problem direct futur work mine databas structur build data qualiti browser abstract data mine research typic assum data analyz identifi gather clean process conveni form data mine tool greatli enhanc abil analyst make datadriven discoveri time spent perform analysi spent data identif gather clean process data similarli schema map tool develop autom task legaci feder data sourc purpos assum structur data sourc understood data set feder dozen databas thousand tabl ten thousand field reliabl document primari kei foreign kei develop system bellman perform data mine structur databas paper present techniqu quickli identifi field similar valu identifi join path estim join direct size identifi structur databas result databas structur mine analyst make sens databas content inform prepar data data mine find foreign kei join schema map identifi step prevent databas collaps weight complex data qualiti browser abstract data mine research typic assum data analyz identifi gather clean process conveni form data mine tool greatli enhanc abil analyst make datadriven discoveri time spent perform analysi spent data identif gather clean process data similarli schema map tool develop autom task legaci feder data sourc purpos assum structur data sourc understood data set feder dozen databas thousand tabl ten thousand field reliabl document primari kei foreign kei develop system bellman perform data mine structur databas paper present techniqu quickli identifi field similar valu identifi join path estim join direct size identifi structur databas result databas structur mine analyst make sens databas content inform prepar data data mine find foreign kei join schema map identifi step prevent databas collaps weight complex data qualiti issu servic provis bill hunt glitch massiv time seri data hunt snark find data glitch data mine method data qualiti critic data analysi bad data lead incorrect conclus problem data detect earli time effort spent ingest analyz paper propos data mine techniqu automat detect data problem commonli encount larg multivari data set data mine method ideal purpos design find abnorm pattern larg volum data discuss import type data integr issu demonstr data mine method dataspher set comparison techniqu earlier work detect glitch mimic error condit discuss artifici data introduct corpor base import decis result deriv data segment custom base grow kind product bui correl product bought inf compar massiv high ddimension data set comparison data set reveal great deal inform time vari natur observ process suppos point data set repres custom activ locat dimension space comparison distribut point data set custom activ chang observ period applic includ data integr check unexpect chang data set problem data collect process propos fast inexpens method compar massiv high dimension data set make distribut assumpt method adapt power classic statist complex high dimension data set gener map data set dataspher compar data set compar dataspher dataspher gener pass data set store databas aggregat 