Browse Prior Art Database

Influencers Identification and Priority Ordering in Recruiment networks Disclosure Number: IPCOM000243586D
Publication Date: 2015-Oct-04
Document File: 6 page(s) / 272K

Publishing Venue

The Prior Art Database


A recruiter will have job openings and reaching out to the right candidates with minimal time & effort is always a challenge. Recruiter reaches out to many channels (posting to career sites, job boards, social networks etc...) As hiring the social way is very effective and lucrative with high success rates, many recruiters are building strong network/connections on social networks trying to leverage the social power. Here recruiter will try to spend energies to reach out and attract the right and best talent (candidate) for the listed jobs.

As recruiters have strong networks with hundreds (and even thousands) of connections/contacts in their social network, it's almost impossible to find & post to the right candidates (contacts). Just blindly posting onto all contacts (or onto their walls) is just not feasible.

It's very important that: Either pick the right talent and post to their wall (or communicate) - This is good but still can limit the options as it is restricted to only recruiter's network (contacts) Pick the 'right' contacts, who can in a way 'influence', meaning, 'possibly' may have the right candidates in their network. Example: I (recruiter) have a job opening for a 'business analyst', I have few folks in my network who work in the roles of product management, and most probably, they might have having contacts in their network, who must be interested or working as 'business analyst' roles. Hence I should target these 'product manager' contacts/friends of mine.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 6

Influencers Identification and Priority Ordering in Recruiment networks

Technical Details

We wanted to find influencers in a network who can maximize the reach of the job description to the right set of candidates.

Basic Criteria

 Influencers should maximize the reach-ability to the matching nodes.

 Identify the influencers for the Job description given a set of network properties (Profile of node information and relationship [strength, probability of influence, forwarding capability etc.,] between nodes).

 Influencers must be as close as possible to the Recruiter node.

 Path for forwarding should be defined through nodes that willing to forward rather than weak paths.

Desired Criteria

 Each person is assigned a budget [typically a number] for receiving/forwarding/influencing the set of the resumes.

o For example, in an attempt to forward a job description to a set of people, we do not want to send more than a certain number of job descriptions to a influencer. A professor might not want to be flooded with more than 2 requests.

 Global maximization for spread of influencers- we do not want to choose the same person for many of the job descriptions for a given time frame or given set of Job descriptions.

 Matching candidates should be reached
o For instance JD 1 and JD 2 are available. Say two influencers are available -> I1 and I2 who can reach to maximum number of people for JD1 and JD2.

o The possible combinations are (a) [JD1 - > I1, Jd2 ->I2], (b) [JD2 - > I1, Jd1 ->I2],(c)[JD1, JD2 -> I1, I2]

     o We would like to choose either (a) or (b) in an effort to spread the influencers.
 Incentive mechanism for the forwarding nodes. (Why should the nodes forward the job descriptions otherwise?).

Technical Challenges

 Large graphs - Computationally feasible solutions.

 Large scale; continual growth

 Distributed, organic growth: vertices "decide" who to link to

 Interaction restricted to links

 Mixture of local and long-distance connections

 Abstract notions of distance: geographical, content, social

 To deal with largely unstructured information - cleansing, extracting, converting to structured information etc.,

System Overview


Page 02 of 6

Network Modeling


Page 03 of 6



Page 04 of 6


 Step 1: Collecting complete data (skills, roles, education, experience etc.) from contacts network of a logged-in user.

     - Storage in the form of a directed graph by noting parents - All contacts are modelled as nodes - All connections between contacts are modelled as edges given probability according to their social closeness - Contacts as deep as possible are considered
 Step 2: Identifying exact attributes (skills, roles, education, experience etc.) from desired attribute set.

 Step 3: Matching desired attributes to all perspective candidates based on Similarity algorithm (explained later)

     - Rank ordering the top candidat...