Notes

Inder
04-07-03

cc: Kartavya

cc: Abhi

cc: Vicky

Organisation Chart - Drawing Tools

See enclosed cutting from Express Computer. I am sure there must be hundreds of software products available in the market offering facility for an HR manager to construct his own company’s Organisation Chart.

I would not be surprised if many such software tools are even “freeware”!

I suppose these softwares would permit construction of Org. charts, as per sample enclosed.

Some (software) tools may even permit entering in each “position-slot” information such as:

Joining Date
Birth Date (or Age)
Cadre
Current Salary
Total years of experience
Edu. Qualification
etc. etc.

Each slot/position becomes a kind of “mini-resume” / “snapshot-at-a-glance”.

Then there could (conceivably) be such Org. charting tools, where, when you move your cursor over the “Designation / Position-Name”, a window opens up displaying a complete Job-Description for that position!

If a company is systematically organised & has already created a Manual of Job Descriptions for all positions in the company, then, to provide such a “mouse-over” feature is not difficult at all!

Our (long-term) objective should be to find (or develop in-house) such an Org. charting software & integrate it with our ResumAct Compose.

If we succeed in this (—and we must), then, everytime/anytime an executive resigns/retires and need arises to fill that slot/vacancy,

All that the Recruitment Manager has to do is to open the Org. chart and double-click on the “position-slot/block”!

And presto!

Draft advertisement appears before his eyes! — ready for editing.

As simple as that! You cannot get any simpler than this.

Recruitment managers around the world would love this “elegance.”

“No-fuss—recruiting at its best!”

This “vision” should not hold up development of V1.0 of ResumActCompose (for which, I would like to see project-plan in next 2/3 days).

And Inder should continue his R&D along this line (including calling S.K. Consultancy to give us all a demo).

(Signature)
04-07-03

Rahul
New Tool

06/01/08

Resume Rater V3.0

We discussed this yesterday.
It has the potential to become a killer App.
We can develop it using Google Desktop API — in which case Google will host it on their own website! This can popularise RR V3 in a very big way around the world.
I am in two minds — whether to develop RR V3.0 as:

a totally separate version, based on Google Desktop OR
integrate Google Desktop as an additional feature into V2.0, leaving intact all the existing features of V2.0

I think you will be the best person to judge this issue from technology application viewpoint.

From recruiter’s/user’s perspective, it would be better if he has to install only one version, whether he wants to search/create resumes lying in:

a specific folder or
on entire hard disk

Example: two clickable buttons of our main page UI of RR:

Rate resumes lying in any specific folder/file (OPEN)

[File] [Folder]

↓

[Rate]

Rate resumes lying in ALL the files/folders on your hard disk (Needs installation of Google Desktop)

[Rate]

If you do not have Google Desktop installed on your PC, click here to download.

After you have downloaded Google Desktop, it will take from a few minutes to a few hours to index the content of your hard disk. Of course it is a one-time/first-time effort.

Search results should be displayed as per our existing display grid, in the descending order of raw-scores.

One more thought. Today we are dependent upon user resetting the counter, in order for our website counter to change.

How can we eliminate this “dependency”?

Can our counter change as soon as user clicks on some button? e.g.: [SAVE Results]

But then, he may not be online at that time!

Nor can we reduce the “reset-limit” from 1000 to (say) 100!

Chances are, there may be 10,000 resumes on his hard disk — all of which he wants searched/rated — simply because he is too lazy to select any particular folder.

In fact, the very idea (behind integrating Google Desktop in V3.0) came from the fact that a typical recruiter can only remember the folders that he has created during last 10/15 days — at most.

Whereas,
the best resumes may be lying hidden in folders that he created 2 months/4 months/8 months ago!

Who, the hell has time/patience to search thru 59 folders created during last 12 months, to find/select those one/two, which are likely to yield best resumes?

This is simply too much to ask!

This is enormous “process-friction,” which limits the benefits of RR V2.0. This is why we need Google Desktop.

Which means, even the current “reset” limit of 1000 may not work! Will not work!

We will need to raise it to 50,000!

We just don’t know how many resumes are lying on the hard disks of a WIPRO/Infosys/IBM/Accenture recruiter!

Of course, I don’t expect “rating process” time to increase appreciably because, like ISYS, Google Desktop has already indexed all documents — and against each document, created a database of several hundred “words”.

Altogether several million — or a billion — words.

So finding takes a few seconds at most (see how fast ISYS does it).

Maybe we can employ “reset-counter”

…only when a user uses specific folder to rate.

…and use some automatic (counter update) process when he uses Rate entire Harddisk option.

Another thing.

In RR V3.0, can we “harvest” entire resume, instead of merely harvesting email ID?

Of course, we don’t want to harvest all 50,000 resumes on hard disks again and again 10 times in a day, if a user rates entire harddisk 10 times a day, for (in sequence):

Java
Asp
VB
AJAX
Windows

etc. etc.

But, conceivably, during (or at the end of) each rating process, we could harvest only those resumes which scored ≥ 60.

This will eliminate most of “duplication”.

Obviously, during each rating, the number of resumes scoring ≥ 60 will vary from a few to a few dozen — which is ok.

We will create, on our webserver:

86 IT skill folders
43 Non-IT skill folders

and deposit/store the harvested resumes into respective folders.

This way, we will create a huge database of “scored” resumes (in descending order too) for all functions and all skills.

(Signed)
08/01/08

Jobsites
Indianfresher.com

Google Desktop

http://code.google.com/?utm_source=en-cpp
Developer Home
Developer Resources
Google Chart API
Google Desktop SDK

Google code home → Google Desktop SDK

Gadget API
Search API

Deepa/Abhi
21-02-04

Searching for Executive Names using HARVESTER / GOOGLE Combination

We have concluded that maybe 70%–80% of the sentences contain:

says
said
according to
“ ” (quote marks)

But, from enclosed write-up, I found some more “keywords” which you may experiment with and, if found productive, may want to add into the “Keyword Dropdown List” in Harvester. These are:

Lost
quit
put in papers / put in his papers
take over as
with (his) resignation
left
(has/have) moved on
(has/have) joined
has been appointed as
(He) will take over
will report to
replaced
joined

(Signed)

Regi/Abhi

06-02-04

“Harvester”

Please file this note in the file given to you. I have no copy.
Enclosed find:

➔ A few “Express Computer” copies in which I have highlighted the executive names. I have also circled the keywords which appear somewhere in the sentence which contains the executive’s name.

➔ Compilation of these keywords. Please remember that this compilation is only for Express Computer. In the folder given to you earlier, there are similar compilations for Business World / Business India etc.

If we plan/intend to carry out the actual “extraction”/“harvesting” ONLINE, then “time-to-harvest” becomes crucial. In this case, we need to have a relatively short list of keywords (to match & highlight the relevant sentence).

In this case, it would be best to have “CUSTOMIZED” keyword shortlists, one each for each newspaper/magazine. The journalists have their own style/preferences of writing. Each journalist has his own “pattern.”

Hence, in a given newspaper or magazine, one PATTERN of keyword usage gets established, which generally does not change frequently.

So a customized list of keywords would take less time to locate & highlight the relevant sentences.

This means, depending upon the newspaper/magazine chosen by Vital/Samanta, the software will automatically select the RELEVANT/CUSTOMIZED list of keywords for matchmaking.

If we were to download pages/articles first and then do the harvesting OFFLINE, then we could afford to “merge” all individual keyword lists of all newspapers/magazines and have one/large unified list. Of course, this would drastically slowdown the harvesting but then it may not matter since we would be doing the harvesting OFFLINE.

We may, eventually, develop a…

Webspider, which simply keeps visiting one website after another (based on a predefined list given by us) and keeps downloading all the pages. This download activity can go on, on one machine, around the clock.

On a second machine, harvesting can go on simultaneously but OFFLINE. So, in this scenario of split-working, time of harvesting is not that critical. On the other hand, by having a SINGLE/LARGE list of keywords, we would achieve a high level of extraction accuracy.

Our eventual goal would be to develop a COMPOSITE/COMBINED “SPIDER-CUM-HARVESTER”, which would visit the pre-decided (pre-configured) websites, locate the “executive name/designation/company name” and automatically create a database without any human intervention.

But we must take one-step at a time. Let us first come out with Version 1.0 as fast as possible so that we can improve our productivity (of manual marking in magazines) as also productivity of Vittal/Samta!

In any case, as discussed yesterday, I would like you to capture/store the relevant sentence (in which the name of executive appears) against the name of the executive. The sentence must also carry the “Source Name” & the (system) date.

Idea is that, if a given executive’s name appears in 4 different magazines at 4 different times (over a period of one year, or any length of time), then by clicking on the name of the executive, one should be able to see/bring all the sentences from which his name was picked out.
This way, the consultant would get a HISTORICAL PERSPECTIVE/OVERVIEW of the concerned executive. This would tremendously improve the probability of our being able to find and recommend to the client the most appropriate candidate. Such an automatic compilation of the background of a given candidate would help to check out his antecedents with his previous employers. This becomes a very valuable input into our headhunting process.

(Signed)
06/02/04

HARVESTER
Extraction Logic for “Express Computer” 06-02-04

Says ___ / Says that ___

According to ___ / according to ___

Said ___ / said ___ / said that ___

Believes that ___

Warns that ___

Agrees that ___ / agrees that ___

States that ___ / states ___

Asserts that ___

Believes that ___

Adds ___