Re:Entity Resolution: The most common data science challenge
by jandrewrogers on Wednesday June 01, 2022 @02:44PM
(#31586304)
So much this. I work on spatiotemporal entity resolution and it requires extreme care and cleverness to not turn it into a proverbial "boiling the ocean" compute problem for all but the most trivial cases. The implied state that needs to be managed alone is often effectively intractable. At one time supercomputers were used for large-scale entity resolution problems; they may still.
[ Parent
| on HN ]
Re:Entity Resolution: The most common data science challenge
by code_biologist on Wednesday June 01, 2022 @10:57PM
(#31590771)
The beautiful thing about spatiotemporal entities is that physical time provides "out of the box common sense" decent locality, and physical space excellent locality, for the blocking process your parent comment mentions.
It sucks to work on a whole product catalog or rolodex.