BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Bjørn Kjos-Hanssen (University of Hawaii at Manoa) DTSTART:20201117T230000Z DTEND:20201118T000000Z DTSTAMP:20260423T005743Z UID:CTA/36 DESCRIPTION:Title: A f amily of metrics connecting Jaccard distance to normalized information dis tance\nby Bjørn Kjos-Hanssen (University of Hawaii at Manoa) as part of Computability theory and applications\n\n\nAbstract\nDistance metrics a re used in a wide variety of scientific contexts. In a 2001 paper in the j ournal Bioinformatics\, M. Li\, Badger\, Chen\, Kwong\, and Kearney introd uced an information-based sequence distance. It is analogous to the famous Jaccard distance on finite sets. Soon thereafter\, M. Li\, Chen\, X. Li\, Ma and Vitányi (2004) rejected that distance in favor of what they call the normalized information distance (NID). Raff and Nicholas (2017) propos ed a return to the Bioinformatics distance\, based on further application- informed criteria.\n\nWe attempt to shed some light on this "dispute" by s howing that the Jaccard distance and the NID analogue form the extreme poi nts of the set of metrics within a family of semimetrics studied by Jimén ez\, Becerra\, and Gelbukh (2013).\n\nThe NID is based on Kolmogorov compl exity\, and Terwijn\, Torenvliet and Vitányi (2011) showed that it is nei ther upper semicomputable nor lower semicomputable. Our result gives a 2-d imensional family including the NID as an extreme point. It would be inter esting to know if any of these functions are semicomputable.\n LOCATION:https://researchseminars.org/talk/CTA/36/ END:VEVENT END:VCALENDAR