JOB PREFERENCES

Abhi
15-06-04

Job Preference for a given “INDUSTRY” – NAME

On last Sunday, we came to a conclusion that:

• “Skills” required for some “functions” such as HR / Legal / Finance / Admin etc. are more or less SAME / IDENTICAL across all industries.

For an executive looking for an opening in such functions, industry does not pose much of a problem. He can succeed in any industry.

For a person working in such a function, as far as “Job Preferences” are concerned, the following parameters might suffice:

• Posting City / Location
• Function (i.e., functional skills)
• Designation (Actual or “Level”)
• Salary offered
• Employer Company Name (because of Reputation)
• Then there are a whole lot of “functions,” where the “functional skills” are really not transferable from one industry to another industry.

Examples:

• Mfg / Production
• Sales / Mktg / Service
• Materials / Purchase / Stores / Supply Chain
• Design / R&D
• Project Mgmt
• Maintenance
  etc.

These are functions where success depends on prior intensive knowledge (and some function-related skills) of:

• “Products” being manufactured
• “Services” rendered (e.g., Consultancy)
• Processes employed
• People to be managed / supervised (skills)
• Customers
• Suppliers / ancillaries
• Materials / components that go into products
• Machinery installed / used
• Info / Data / Knowledge etc.

By working in a given industry, an executive acquires such knowledge/skills (relevant/pertinent to that industry), over a long period of time.

And such “knowledge/skills” differ from one industry to another industry. These are not easily “transferable.”

---

And this “Locked-In” nature of knowledge/skills is the main reason why:

• Corporates quite often mention in job-adverts that candidates must have prior experience in such & such industry.
• Jobseekers quite often don’t even search/look at job-adverts from any industry other than their own! They know that if they were to apply to any other industry, then:
• In first place → they will get rejected.
• In second place → If, by fluke, they get selected, they will make a fool of themselves, thus “guaranteed” failure!

If corporates (job-adverts) do not state explicitly that the candidate must be from XYZ industry, then they make this implicit (indirectly clear) by listing a dozen “skills,” which you can expect to find only in that industry!

From this, we can conclude that:

• Each industry has its own set of unique:
• Knowledges
• Skills
  Which are a must for working in that industry.

• In a job-advert, such “industry-specific” knowledge/skills are described thru some KEYWORDS.
• Such “keywords” describe: products / services / processes / people / customers / suppliers / materials / components / machinery, etc. which are unique to that industry.

Occasionally, there will be some words which are common amongst a “set/group” of industries, which are similar.

Occasionally, same keyword might be found in totally different industries – but with lower/different “frequency of usage”. This is our experience in case of “functional” keywords also.

• Most job-adverts do not say that the advertiser-company belongs to XYZ industry!
• And even if it says so, their “Industry Name” may not correspond to our naming convention.
• Even if we employ dozens of experts, it would be impossible to “map” company names with industry names, considering that there are … [text continues on next page].

Lakhs & lakhs of companies around the world, whose job-adverts we intend to push onto JAM-subscribers’ mobiles. Even within India, there are hundreds of companies with sales turnovers in excess of Rs. 100 crores, whose names even we have never heard of!

So, “mapping” is out of question.

• Altogether “new” industries are being born and “old” industries keep dying. So, keeping same set of industry-names also won’t work forever.

This means, we have to work backwards.

• First, find/index keywords found in a job-advt. Then
• Based on our “finding,” determine the industry-name of the job-advertiser.
  (I presume this process can be offline).

What we need is a starting database, which fortunately we have!

Some 2 years back, I believe we got Rajiv to develop a software tool, and loaded it with 1000 job-adverts.

Then, we asked Mr. Anjara to highlight from each job-advt:

• Function-related keywords
• Designation-level related keywords, and
• Industry-related keywords.

And then decide/determine that this particular job-advt belongs to:

• Function = Sales
• Designation Level = Manager
• Industry = Chemicals

In fact, the tool looked like this (sketch):

Advt. No.  |  Description

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

This Ad belongs to:

Function ▼   Level ▼   Industry ▼

Below it showed reasoning:

• Functional keywords found: ____
• Designation-level keywords found: ____
• Industry-related keywords found: ____

Anjaria analysed 1000 job-adverts (for which we paid him Rs. 10/- per advt.) in one month.

But since sample was small (only 1000 ads) and maybe, all 50 odd industries were not uniformly represented/distributed (i.e. 20 each), the database created never got used in GumMine/GumSearch.

But now, if we take 200 job-adverts from each of the 50 industries, and then get Anjaria/Inder to “process” these 10,000 job-adverts, then we would build up, for each industry, a fairly representative database of:

• Knowledge keywords
• Skills
• Product/Process
• Education/need
• Materials
  
  

Then we calculate the “frequency” of usage of each keyword for a given industry & assign probability.

To begin with, even if we find (and arrange in descending order of usage), say, only 100 keywords in each industry, that would make a good beginning.

These 100 keywords (per industry), in …

My opinion, would be good enough to identify a new/incoming job-advt with 80% accuracy as belonging to XYZ/ABC industry.

And then, of course, some job-adverts will bring in new/additional keywords, which now our software must automatically “add-up” to the Industry-wise Keywords Databases.

And thereafter keep calculating “frequency of usage” every time that keyword appears in any job-advt.

So, as in case of keywords for functions in GumMine, keywords for industries in JobMine would also become a self-learning, automatic process!

And, if I am not mistaken, this self-learning process has already been made a component in GumMine’s .NET Architecture!

Same can be plugged into JobMine!!

So, all that is required to kick-start the process is to:

• Find 200 x 50 = 10,000 job-adverts
• Get hold of Anjaria & Inder for 5 months
• Invest Rs. 1 lakh (10,000 x 10)

We should be willing to risk this amount, provided you agree that we are on right track – assuming of course that there is no better/faster/cheaper process!

I await your feedback/recommendation.

(Signed)
15/06/04

---

P.S.
In enclosed “Times Ascent,” see the keywords which I highlighted & having done, reached a “conclusion” re: the “Industry” of job-advt. Anjaria/Inder would do so, on computer.

 Abhi

19-06-04

The Power of Synonyms

(How to build a Job-Preference Profile)

Everyone does it – eBay / Amazon / Google / Microsoft etc. All search engines rely on this.

And this is why, when you select a book (just to read the preface, not to purchase), Amazon search engine “recommends” to you some other “related” books.

“Related” in terms of:

• Author (books by same author)
• Subject/Topic (books written by others on same/similar topic)

We too, will apply this technique in recommending jobs to candidates – which, we have concluded (statistically), might be of interest to him.

We will do it in 2 stages.

Preliminary Stage

This is the stage when we will “gather” data about a candidate’s “explicit” job-preferences – preferences which he will “spell-out” to us, in so many clear words.

So at this stage, he will be asked to “narrow down” his choices by selecting appropriate:

• Posting location (geographic preference)
• Functions
• Designation levels

Now, out of these, I have suggested for FUNCTION:

(sketch showing a dropdown for Function and a box for highlighting keywords within the function selected)

Here we will let him:

• Highlight those “keywords” which appeal to him
• Allow him to type in additional keywords

What else can we do?

We can offer him “synonyms” subspace.

He selects as his “Function” choice:

Marketing → Keywords to Highlight

We can flash a message:

“Would you be interested in…?”

Example:

• Sales → Keywords to highlight
• After Sales Service → Keywords to highlight

Why?
Because these “functions” have something “common” – some similarity – some overlap.

This can also be done in reverse.

Suppose:
Marketing → Keywords highlighted (Sales, Service, Dealer Network)

[Diagram with overlapping circles]

• Marketing → Sales → After Sales Service → Advertising → Public Relation
• "Common Keywords" in overlapping shaded areas

---

and so on and so forth!

Like in WWW.VisualResaums.com!

This way a candidate is enabled/empowered to put together his own "thali", from different dishes in an
"A-LA-CARTE" fashion (as per card).

He builds up his UNIQUE menu!

And, we are only giving him "ideas" by displaying / asking, “Is it possible that, here too, there is some keyword (dish) that you like?"

SECONDARY STAGE

In this stage, we will archive (against name of each candidate) ALL the job-adverts where he clicks APPLY ONLINE.

The obvious assumption is that these are the job-adverts which he liked — otherwise, he would not have "clicked"!

So, over a period of time, we would create a "database" of 10 / 100 / 1000 job-adverts against each candidate.

Now all we have to do is:

→ Index all "keywords" in all those adverts (could be done once-a-week)

→ Calculate their "frequency of occurrence" & the probability & arrange these keywords in descending order of their probabilities.

Now we know the "Likes" of that candidate.

As he clicks more & more job-adverts the database of keywords (population) grows.

… grows bigger & bigger. And the “probabilities” become more & more “accurate”!

Now, all that the Matchmaker has to do is to compare the “keywords” contained in any job-advt, with this
JOB PREFERENCE PROFILE

which we have built-up for each & every candidate, over a period of time.

And now, when the Matchmaker finds a job-advt closely resembling/matching a candidate’s Job Preference Profile, that job-advt must get shortlisted & delivered as SMS.

The higher the resemblance/congruence, the better the “Match”.

So, as time goes by, Matchmaker would continuously deliver better & better (more relevant) job-adverts to each candidate!
Matchmaker would even “guess/estimate/predict” accurately:

→ Where (city/region/country) does this candidate like to work

→ For which “companies” does he like to work

→ In which “functions” does he like to work

→ At what “designation-level” does he like to work (this will of course shift upward with passage of time)
→ What actual “position/vacancy” names does he think attractive, etc. etc.

Our Matchmaker Software must have this built-in provision (and right from day one – i.e. NOW), to continuously build-up such
JOB PREFERENCE PROFILES

For each & every JAM Subscriber. Building such “Learning” into the software alone can DIFFERENTIATE us from other jobsites, where

→ the jobsite does NOT “remember” anything

→ jobsite makes a candidate enter his “job-search parameters” again & again & again, forever.

Let us be DIFFERENT / SUPERIOR.

(Signed & dated: 13/06/09)

 Dated 11-06-03

Job Preferences

Dear Jobseeker,

We aggregate over 5000 job-adverts daily from a number of jobsites (List).

But, out of these huge numbers, you want us to alert you (as SMS on your mobile phone), about no more than 5 or 10 job-adverts, which are most relevant to you.

You alone can define what jobs are relevant to you by carefully selecting your job-preferences in terms of:

• Geographical Preferences
• Functional Preferences
• Designational-level Preferences
• Vacancy/Position Preferences

The more sharply you define your preferences, the better we are in a position to find you your DREAM JOBS!

Although you are welcome to come back anytime and modify/edit your job-preferences (and as often as you wish), please carefully select below,

MY JOB PREFERENCES

Password []  User ID []

[Submit]

▶ Geographical Preference (Mandatory)

• Select COUNTRY
• Any / All
• America
• Brazil
• India
• Select CITY
• Ahmedabad
• Benaras
• Cochin
• Dehradun

Notes:

• If you select “Any/All” for country, then thereafter you cannot select any specific city (– there are thousands of major cities in the world!). In such a case, “Select CITY” drop-list will remain blank.
• But if you select a specific (only one, of course) country, then “Select CITY” drop-list will display 20 major cities of that country, out of which, you can select a maximum of 3 cities.

 

▶ Functional Preference (Mandatory)

Notes:

• You can select only ONE function at a time from the following drop-list (– but of course, you can come back and edit!).
• Selecting any function will display corresponding keywords in adjoining box. Feel free to highlight one or more of these keywords which you expect to be mentioned in the job-adverts.
• Besides highlighting, you can also add/type new keywords in the “Keywords Box.”

[Diagram]

• Function List [dropdown]
• Keywords Box:
• Highlight [], [], [_____]
• Type new ones [___________]

 

Designation-Level Preference (Mandatory)

Notes:

• Not all, but quite a few job-adverts indicate the hierarchy/designation level of the position advertised, even if indirectly.
• Then there is the matter of public perception (e.g.: Is a “Financial Controller” at the same level as V.P. (Finance), who, in turn, is at the same level as a CFO?). Since position-descriptions differ from company to company, even for similar/identical job responsibilities, it becomes necessary for you to tell us at what LEVEL you wish to work.
• So, please select ONE from the following:

[Diagram]

• Designation Level [list with radio buttons]

 

Vacancy / Position Preference (OPTIONAL)

Notes:

• Quite often, Vacancy/Position mentioned on top of a job-advt gives you a fairly good idea of:
• Functional background (desired)
• Designation level (offered)
• Skills (expected)
• Industry background (desired)
• For a great majority of candidates, the “Vacancy/Position” description says everything they want to know before applying.
• Only problem is, there are thousands of “Vacancy/Position” descriptions and we can only show a few below for ENTRY LEVEL / JUNIOR LEVEL POSITIONS (of course, feel free to type any specific “Vacancy” that you cannot find below).

[Diagram]

• ENTRY LEVEL / JUNIOR LEVEL VACANCIES (Select One Only) [dropdown with radio options]
• Type: [____________________]