BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:America/Chicago
BEGIN:DAYLIGHT
DTSTART:20260308T030000
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3
TZNAME:CDT
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20251102T010000
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11
TZNAME:CST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260115T190358Z
UID:74515F61-1F03-448D-A3ED-F295BEDC95BB
DTSTART;TZID=America/Chicago:20260115T120000
DTEND;TZID=America/Chicago:20260115T130000
DESCRIPTION:Entity resolution is the backbone of any data platform that aim
 s to present a single\, trustworthy view of an organization across noisy\,
  overlapping sources. This talk shares a practical\, system-oriented bluep
 rint for companies entity resolution that you can adapt to your stack. We
 ’ll begin with upstream data preparation—standardization\, canonicaliz
 ation\, and normalization of names\, websites\, addresses\, and phones—t
 o reduce ambiguity before matching. We’ll then cover signature construct
 ion (e.g.\, relaxed/collapsed variants)\, blocking to avoid N² explosion\
 , and a match function that combines exact agreement on one core attribute
  (website\, name\, or address) with a second fuzzy signal to balance preci
 sion and recall. You’ll see how constraints (e.g.\, unique primary websi
 te\; unique name+HQ address)\, attribute scoring/selection\, and separatio
 n of company vs. location resolution improve quality and explainability. W
 e’ll discuss pre‑merge signals from authoritative linkages\, human‑i
 n‑the‑loop controls for edge cases\, and governance patterns—provena
 nce (“why this value”)\, rollovers for stable IDs\, and reproducibilit
 y. Finally\, we’ll outline evaluation and monitoring tactics (drift chec
 ks\, audits) and deployment considerations for both batch and streaming en
 vironments. Attendees leave with a clear set of building blocks to move fr
 om noisy inputs to reliable\, auditable entities.\n\nSpeaker(s): Rohit Mut
 hyala\, \n\nVirtual: https://events.vtools.ieee.org/m/520810
LOCATION:Virtual: https://events.vtools.ieee.org/m/520810
ORGANIZER:bishopm@acm.org
SEQUENCE:3
SUMMARY:Computer Society Webinar - Large scale Entity Resolution: A Practic
 al Blueprint from Noisy Records to Trustworthy Entities
URL;VALUE=URI:https://events.vtools.ieee.org/m/520810
X-ALT-DESC:Description: <br /><p>Entity resolution is the backbone of any d
 ata platform that aims to present a single\, trustworthy view of an organi
 zation across noisy\, overlapping sources. This talk shares a practical\, 
 system-oriented blueprint for companies entity resolution that you can ada
 pt to your stack. We&rsquo\;ll begin with upstream data preparation&mdash\
 ;standardization\, canonicalization\, and normalization of names\, website
 s\, addresses\, and phones&mdash\;to reduce ambiguity before matching. We&
 rsquo\;ll then cover signature construction (e.g.\, relaxed/collapsed vari
 ants)\, blocking to avoid N&sup2\; explosion\, and a match function that c
 ombines exact agreement on one core attribute (website\, name\, or address
 ) with a second fuzzy signal to balance precision and recall. You&rsquo\;l
 l see how constraints (e.g.\, unique primary website\; unique name+HQ addr
 ess)\, attribute scoring/selection\, and separation of company vs. locatio
 n resolution improve quality and explainability. We&rsquo\;ll discuss pre
 ‑merge signals from authoritative linkages\, human‑in‑the‑loop con
 trols for edge cases\, and governance patterns&mdash\;provenance (&ldquo\;
 why this value&rdquo\;)\, rollovers for stable IDs\, and reproducibility. 
 Finally\, we&rsquo\;ll outline evaluation and monitoring tactics (drift ch
 ecks\, audits) and deployment considerations for both batch and streaming 
 environments. Attendees leave with a clear set of building blocks to move 
 from noisy inputs to reliable\, auditable entities.</p>
END:VEVENT
END:VCALENDAR