June 13th release of WebVeta had a private preview of A.I based search. But, here is the question? How do I measure the quality of results? There needs to be a way for measuring the quality of the results and this metric needs to be improved!
My special thanks and gratitude for Microsooft for Startup Founders, NASSCOM 10,000, Codeium.
https://foundershub.startups.microsoft.com/signup
https://10000startups.com/virtual-incubate-program
Customers, who sign-up prior to 30/06/2024 get unlimited access to free features, newer features (with some restrictions), but for free for at least 1 year.
I have used a small sub-set (1000 rows) of a large publicly available database containing over 2.3m rows.
The following are based on 1000 random different texts, the accuracy might vary among datasets, but hey this is a starting point.
The base metrics were:
60% of the times the first result was relevant.
70% of the times the relevant results are in the first page i.e first 10 results.
Certain tweaks were made and much more tweaks are planned. But the results of the past 24 - 48 hours:
67% of the times the first result was relevant - I would like this metric at 90% or more.
89.5% of the times the first 10 results had the relevant result - I want this metric at 100%
From speed optimizations perspective, small VM had these approximate numbers:
Base - approximate 200ms per search
New - approximate 40ms per search
Ingesting data entirely different because for most content, ingestion happens once but retrieving needs to happen fast and probably several thousands or several million times. At this point, not much concerned about ingestion speed, but probably later.
In other words the past 24 - 48 hours had 500% speedup, 11.67% increased accuracy in 1st answer, slightly over 27% increased accuracy in 1st 10 results.
Over the next few weeks, higher accuracy A.I based search, general availabillity of A.I features and paid tier are going to be introduced.
I don’t have any fake aliases, nor any virtual aliases like some of the the psycho spy R&AW traitors of India. NOT associated with the “ass”, “es”, “eka”, “ok”, “okay”, “is”, erra / yerra karan, kamalakar, diwakar, kareem, karan, erra / yerra sowmya, erra / yerra, zinnabathuni, bojja srinivas (was a friend and batchmate 1998 – 2002, not anymore – if he joined Mafia), mukesh golla (was a friend and classmate 1998 – 2002, if he joined Mafia), erra, erra, thota veera, uttam’s, bandhavi’s, bhattaru’s, thota’s, bojja’s, bhattaru’s or Arumilli srinivas or Arumilli uttam(may be they are part of a different Arumilli family – not my Arumilli family).
Mr. Kanti Kalyan Arumilli
B.Tech, M.B.A
https://www.facebook.com/kanti.arumilli
https://www.linkedin.com/in/kanti-kalyan-arumilli/
https://www.threads.net/@kantiarumilli
https://www.instagram.com/kantiarumilli/
https://www.youtube.com/@alighttechnologyandservicesltd
https://www.youtube.com/@kantikalyanarumilli
https://twitter.com/KantiKalyanA/
https://kantikalyan.medium.com/
https://www.facebook.com/ALightTechnologyAndServicesLimited/
https://www.linkedin.com/company/alight-technology-and-services-limited/
Founder & CEO, Lead Full-Stack .Net developer
ALight Technology And Services Limited
Phone / SMS / WhatsApp on the following 3 numbers:
+91-789-362-6688, +1-480-347-6849, +44-07718-273-964
+44-33-3303-1284 (Preferred number if calling from U.K, No WhatsApp)
No comments:
Post a Comment