Rajeev
cc: Vikram cc: Rahul
24-02-06
Office
On the Go
▶ Read enclosed cuttings
▶ In my notes (Job Parade
/ Multiplexes / Breaking News), I spoke about making available such RSS
feeds to websites / portals, etc. (other than job sites).
Diagram
Center:
Glo. Rec. (Global Recruiter)
Feeds
to:
- News / Papers: JAS
(Already implemented)
- Mobile Screen: JAM
(Already implemented)
- TV Screen: JAT (Job Market on TV)
(To be implemented)
- PC Screen: Job Parade / Multiplexes /
Breaking News (To be implemented)
Note:
Database remains same; only Delivery Channel changes.
Depending upon screen size, presentation will change.
|
Screen |
Level
of Interactivity |
|
1.
PC Screen (Job Parade, etc.) |
•
Highest interactivity• Can view full job advert• Can immediately apply
online• Can see latest jobs fading in & fading out (drama)(Reach:
30/40 million) |
|
2.
TV Screen (JAT) |
•
Little less interactivity• Can see job summary or streamer at bottom of
screen• Note down job code & apply online from mobile• In future – with
TV remote(Reach: 80/100 million) |
|
If
IPTV |
(arrow
from above) – possible higher integration later |
|
3.
Mobile Screen (JAM) |
•
Still smaller screen, so 160-character SMS• Can apply online(Already 80
million, growing 5 million/month) |
|
4.
Newspapers (JAS) |
•
None (no interactivity)• Must note down Job Code & then use “Resume
Counter”(Reach: 400 million readers) |
Rahul
cc: Vikram cc: Rajeev
24-02-06
Inbox
Reader – Automation of Email Sorting
- About two years back, Vikram developed a
tool — which we called “Inbox Reader”, which I believe Vrital is
still using today — to automatically separate out:
- “Resume” emails
- “Correspondence”
emails
From
all incoming emails, “Inbox Reader” figures out which ones are Resumes
and then puts those in a separate folder.
Earlier,
this process was being carried out manually by a girl, and used to take a full
day!
Today, “Inbox Reader” takes only a few minutes — a tremendous increase in
productivity.
Human
Use of Human Beings!
- You can find out from Vikram what logic
he had used in developing this tool.
- Most spam-filter softwares are based on
the Bayesian Theorem of Probability.
Many are free downloads. After checking with Reema, download some and experiment.
Check how good these are at learning as more and more usage takes place.
Page
2
But
the s/w learns from human beings.
Human has to start off by telling the s/w:
If
you come across the following words in the email, treat that email as spam and
place it in a separate folder.
This
is the starting point — the s/w looks for these “words” in the email.
But
some emails which are really spam, yet do not contain any of the bad words,
will escape.
So,
the human will carefully look at such “escaped” emails and try to figure out
some more bad words.
Then add these new bad words to the list.
Now
the s/w has a bigger knowledge base.
This process goes on, improving accuracy of detection and elimination.
Keeps
improving.
GunMine
works somewhat similarly.
For each “Function”, we have given it a starting Knowledge Base of 30–40
keywords.
If
it finds these keywords in a text resume, it concludes that:
- ABC resume belongs to “Sales”
- XYZ resume belongs to “Mktg”
Except:
In GunMine, each keyword also carries its past frequency of usage
(Weightage).
So keywords found are arranged in descending order of weightages.
You
should thoroughly understand (from Vikram / Rajeev) how GunMine / GunSearch
work.
(Signed)
24/02/06
“Replace
Old Currency with New Currency”)
Rajeev
cc: Vikram cc: Rahul
24-02-06
Replace
Old Currency with New Currency
(text
resume → ImageBuilder)
- It is for this reason that we have
decided to give away (free) GunMine & GunSearch (without
the Knowledge Base) to subscribers.
- For the same reason, we must allow /
enable a jobseeker to download his ImageBuilder (including Function
Profile Graphs) as soon as he posts his text resume during the Submit
Resume process.
- Once he has downloaded ImageBuilder, we
expect that he will in future use only ImageBuilder whenever he
applies offline.
But
then, he may not use it if his raw score / percentile score is poor!
Such jobseekers will continue to send their plain text resumes to HR managers.
By
the same logic, those candidates whose raw score / percentile score are very
good would certainly like to impress HR managers by sending their ImageBuilders,
even offline.
I
believe these candidates will keep coming back again and again to download
their latest copy of ImageBuilders—containing their latest GRAPHS with
updated raw score / percentile score.
They
would be keen to find out whether their score is getting better or worse!
(Of course, the raw score won’t change unless they edit / modify their text—by
adding some more keywords.)
But
as more and more executives belonging to the same Function start
depositing their resumes, then the percentile score of everybody will
change.
This
effect is quite like my looking up my L&T stock price every day in
the evening (the only stock I own)!
Not a day will pass without seeing this.
Most
people who own several stocks look up newspapers first thing in the morning and
TV channels in the evening.
Some even get live stock quotes on their mobiles several times a day!
All
depends upon your level of investment — the greater the investment, the greater
the involvement / intensity.
Then
CFOs, MDs, Chairmen of each listed company never fail to look up their own
company’s stock price several times a day — their jobs literally depend upon
their share price!
Thus,
through introduction of ImageBuilder, we have a similar idea.
The Function
Profile Percentile Score is a kind of a professional’s
“Personal
Stock Price.”
Especially
if HR managers attach a lot of importance to this score.
Then
you would want to come back to Global Recruiter every day and watch its
movement!
Imagine
if 3 / 5 lakh jobseekers were to log in every day just to watch their latest
percentile score —
That
would make Global Recruiter the most sticky website in the world —
where a person must return every day!
Let
us work towards this.
(Signed)
24-02-06
No comments:
Post a Comment