Notes: SEARCH ENGINE

CLIENT | | PROFILE | CANDIDATE - DATA.

| | GENERATOR. |

| | PERSONAL DATA: |

Telephonic | Edu. Quali | Name | - Typed

Requests | occasionally | (Name of Company) | Addresses | Bio-data

(MANUAL) | | Edu. Qualification |

| | Current Employer |

Executive | Experience. | Office Phone No | - Handwritten

Search | ✓ (Level, | Resr. | Executive

Request | Range. | Past Employers | Data Sheet

(ESR) | | Periods Worked. |

| | Designations |

(Individual) | | ✓ | KEY-WORDS |

| | ✓ | Industries |

ADVT | | ✓ | Functions |

SUMMARIES | | ✓ | Products/Services |

| | skills |

Software) | | | |

Entry | | | | Executives

| | | | on | Overseas

| | | | 11-1-97. |

Proposed Layout for Storage of scanned/AI data into an intermediate file

1. Name of Table:

HHTBL1.DBF

2. Contents:

To store data picked up from scanned bio-datas.

3. Layout:

Details Field Name Num/Chr Len Dec

----------------------------------------- ---------- ------- --- ---

Record Name(Fixed 'HT1') Recname C 3

Library Sr. No.(Fixed 201) Libno N 3

Record Status (Leave blank) Recstat C 1

Executive Sr. No. Exsrno N 8

Sub No.(by default 1) Subno N 3

Type of data stored Typdata C 10

Code related to data(if any,) Typcode C 10

Actual description related to this

data element Text1 C 30

Any Numeric Data Numdata N 12 4

----------------------------------------- ---------- ------- --- ---

Year of Passing an exam.(for edun.only) Passyr N 4

----------------------------------------- ---------- ------- --- ---

Speak/Read/Write-Y?(For Language only) SRW C 3

(store Y else blank e.g. Y Y)

----------------------------------------- ---------- ------- --- ---

Current Employment-Y?

4. Type of data stored:

Use following codes for individual data-

Typ data Details

-------- --------------------------

SOURCE Source of Application

ADREF Admt. Ref.

BOXNO Box No.

ESNAME Executive's Surname

EFNAME First Name

EMNAME Middle Name

ESEX Sex

EDOB Date of Birth stored as dd/mm/yyyy

EADDR1 Address Line 1

EADDR2 -2

EADDR3 -3

EADDR4 -4

ECIT City of Residential address

EPIN Pin code

EPHONE Phone nos. separated by ,

EFAX Fax no.

EHOUSTYP Type of House C-ompany flat/O-wn

4. Type of data stored(contd....):

Typ data Details

-------- -------------------------------------------------

ESAL Executive's Current Salary-Basic + DA

ECALLOW -Allowances

ECPERKS -Perks

ECREIMB -Reimbursements

ECANNUAL -Annual benefits

ECRETIRE -Retirement benefits

ECOTHER -Other earnings

ECGROS -Gross salary/Annum

ECRATING -BGC etc. Rating

EDUNCD Education qualification code

PASSYR Year of passing the exam

UNIV University from where passed

LANGCD Language known code

SPEAKY Y, if can speak

READY Y, if can read

WRITEY Y, if can write

EMCONN Organisation in which employed

DESGNN Designation

DESGLV Level of designation (between 10 to 90)

STARTM Employment started during yyyy-mm

UPTOYM Employment ended during yyyy-mm

DURYYMM Period of employment in terms of years & months

STAFNO No. of staff members reporting

RESPON Responsibilities(The activity.... details)

REPOTG Reporting to whom(say M.D.)

WHYLEFT Reason for leaving(descriptive)

MEMBORG Membership organisation

MEMBDESG Designation held there

MEMBPERI Period of membership

REFNM Reference Person Details - Name

REFDESG - Designation

REFADDR1 (-----Residence----) - Address Line-1

REFADDR2 - Address Line-2

REFADDR3 - Address Line-3

REFADDR4 - Address Line-4

REFPIN - Pin code

REFPHONE - Phone Nos. separated by ,

REFOADDR1 (-----Office-----) - Address Line-1

REFOADDR2 - Address Line-2

REFOADDR3 - Address Line-3

REFOADDR4 - Address Line-4

REFOPIN - Pin code

REFOPHONE - Phone Nos. separated by ,

INDUST Code of Industry ) Search Parameters

WORKFN Code of Function )

PRODLN Code for Production Line handled )

CITY Code for Preferred City )

KEYW Keyword data

PRCONM Name of Preferred Company )

COURNNM Name of Course attended

COURDUR Duration of Course

COUNTRY Country in which the course was attended

VCONTRY Code of Country visited (as per postal code 91-India)

VPERIOD Period of visit

INPUT

SCANNED BIODATA

INPUT

VOICE

PHONE-IN

BIODATAS.

STEP #2

OS/2/WARP 4

Voice

Speech

Recognition

STEP #1

(implemented)

ASCH

TEXT

WORD-

FILE

Creation

Bio-Datas

Text form

STEP #3.

FLOPPY

INPUT

EDS on

Floppy

STEP #4

INPUT

MTNL

VSNL

INTERNET

MTNL/VSNL

INTERNET

REMOTE LOG-IN

BY CANDIDATE

STEP #5

ARDIS

DECIPHERING

WORD

RECOGNITION

Each key-word

put in respective

slots

(Temporary Data Storage)

MODULE

CLIENT

REMOTE

CANDIDATE

REMOTE

3 P

LOCAL

FAX

PRINTER

E-MAIL

ARGIS

CONVERTED BIODATA

HARMONISED

DATA-BASE

MTNL

OS/2/WARP 4

VOICE

COMMAND

MTNL/MODEM

SOFTWARE

COMMANDS

2-1-1997.

STEP # Target/Date.

Implemented.

1 Scan bio-datas & Convert to ASCII text to create "WORDS"

2 Install OS/2/WARP 4 speech-recognition software and link it to ASCII Text File (PHONE IN BIODATA)

3. Create Screens (with instructions-MENU) on floppy containing all EDS information for direct transfer to TEMPORARY DATA STORAGE

4 Create Software and physical link which will enable candidates to download (remotely) EDS "Screens", then enter at his leisure & when ready - upload (again by remote log-in) into our TEMPORARY DATA STORAGE. This will be a dial-up MODEM link especially for software professionals in SEEPZ.

For all others who do not have a ready access to a computer-modem link OR who are not COMPUTER-LITERATE, we must immediately plan

VOICE INPUT (OF BIO-DATA)

thru plain old telephone

and

convert to ASCII then OS/2/WARP 4.

In this case, literally LAKHS & LAKHS of executives can get hold of a phone (even at midnight) and dictate

HOW TO DETERMINE/DECIDE/DECIPHER

"RESPONSIBILITY-RELATED" PHRASES/

SENTENCES?

A. One simple criteria is

"any set of words/phrases/sentences or

even paragraphs that follow

- I am/was responsible for - - -"

B. If not, an indirect method is to

examine "clues".

These "clues" are some VERBS/

ADVERBS with or without a

preposition attached.

Some examples appear on

the following pages:

Copy given to Hugh on

4-1-97

31-12-96

WORDS WITH PREPOSITIONS

* In charge of

* Monitoring of

* Comply with

* Filling of

* Maintenance of

* Conversant with

* Finalization in

* Involved in

* Working Capital management through

* Fixing of

* Revising of

* Preparation of

* Comparison of

* Analysis of

* Controlling of

* Guidance on

* Operations of

* Annual Sales of

* Sales of

* Coordination with

* Designing of

* Presentation of

* Making of

* Balancing of

* Finalization of

* Selection of

* Developing of

* Upgrading of

* Imparting of

* Holding of

* Implementation of

* Maintaining of

* Organising for

* Arranging of

* Interfacing with

* Management of

* Liason with

* Involvement in

* Optimisation of

* Control of

WORDS WITHOUT A PREPOSITION

* Managed - / Managing

* Project management which involves - -

* Arrange

* Monitoring. - Research

* Controlling - Formulating

* Identified - Implementing

* Selected - Planning

* Made - Establish

* Computed - Create

* Erased - Develop

* Developed - Expand

* Set-up - Structure

* To enhance - Evolve

* Marketing - Dev Improve

* Conceiving - Widen

* Established

* Upholding

* Steering

* Designing

* Visualizing

* Liasoning

* Overseeing

* Monitoring

* Inducting

* Developing

* Formalizing

* optimizing

* Consolidation

* Build-Up

* Motivate

* Achieve

RESPONSIBILITY-RELATED WORDS

WHICH ARE

- neither ending with "ing" or "ed"

- nor accompanied by a "preposition"

but

- are "stand-alone" (nouns ?)

* Corporate Relations

* Communications

* Press & Media relations

* Legal Affairs

* Administration

* Foreign Collaborations

* Technology transfer

* Joint Ventures

* Capital Goods Import

* Capital Issues (CDR, FCR, ECB)

* Forex Approval

* Public Relations

* Export documentation

* Bank Negotiations

* Licenses / Licensing

* Insurance

* ECRC

27-12-96

Dear Hugh:

In my note dated 23rd inst, I have talked about "CATEGORIES OF WORDS".

For each of this category, we have to build-up a **DIRECTORY** containing hundreds or thousands of words.

Under "**MANUFACTURING-RELATED**" category, there are following sub-categories:

* Equipment – Machinery – Systems

* Raw Materials – Intermediates

* End Products – Services.

I enclose a floppy containing some **1800 + words** that come under above-mentioned categories. Each word has a **FOUR DIGIT** neumantic code as well. These are compiled from **KOMPASS DIRECTORY**. These codes and descriptions are standardised by **UNIDO** or **GATT (WTO)** and are known as **HNS (Harmonised Nomenclature System)** used by custom authorities around the world.

With regards

H. C. Parekh

cc: Mr. Nagle

23-12-96

Dear Hugh:

I hope you had an opportunity to go thru my earlier notes.

As mentioned earlier, there are 2 ways of going about our task of "**deciphering**" scanned bio-datas.

**#1 – STORAGE METHOD**

Create and Store massive **directories** of words in computer-memory. As the software picks-up each word (while scanning/OCR), it compares it with the stored "directories", finds a "**correspondence**" and stores it accordingly.

This is what **RESUMIX** seems to have done with **80,000 words** directory.

In next few pages, I have listed

* categories of words

* some examples.

But, the problem is to manually compile such directories. It may take **months**!

**#2 – [AI METHOD]**

Under this method, we begin with

* a rudimentary "**logic**" (grammar)

* a small list of words' categories

* **theory of probability** (frequency – and by deduction, the probability of occurrence of any given word

* “before” another word

* “after” another word)

The trick is to write a software which is **self-learning**!

It is like providing a small "**nucleus**" around which, gradually a **PEARL** is cultured!

In following pages, I have attempted to establish a **CONTEXT** of words by picking actual examples from 30/40 bio-datas. I am clueless as to how this analysis will help you – but it

may.

The "**word-categories**" mentioned by me are not exhaustive. Please feel free to add.

In the file I sent to you yesterday, I had used following logic:

**Step #1**

Pick a "**word-category**" e.g.

"**Unit | Section | Division | Branch – Group – Dept – Project – Shop – Cell – Area – Office**"

all of which mean "**work location**"

**Step #2**

Find a list of words which fall under this "**word-category**" e.g.

* Tool Room – M/C Shop – Design Cell

* Jig & fixture Assy – Forging Dept – File Production

* Reebok project – Central Marketing Office