An explanation
of document
analysis & matching

 

This page provides more insight in the process of analysis, matching and ranking as perfomed by BeWorkHappy’s technology

1. Analysis


Analysis of relevant concepts
Extraction of one or more domains
Extraction of the senoirity
Extraction of the working class

2. Matching


On document level A job and a resume should:
- have similar domains
- match with respect to seniority
- match with respect to the extracted
working class
On concept level

3. Ranking


For each document that has successfully passed the matching test: Each extracted concept will be assigned to a specific category, which has a predefined weight. The sum of all the concept weights gives the total ranking percentage.

 

Cross language – language independent: all analyses – and matching – is done in a language independent way. This means that BeWorkHappy reads accross all languages it has deployed in the specific solution. For example : English resumes will be perfectly and directly matched with job descriptions in French, and Chinese jobs will find the best-matching resumes of English-speaking candidates …

Regardless whether the database holds 10,000 documents or several millions, our technology always performs this 3 step-process :

1. Analysis of each document.

  1. Analysis of relevant concepts. Example from the vacancy: “should be familiar with a scripting language”,
    “scripting language” has been recognized as a synonym of our concept “Scripting Tools”.
  2. Extraction of one or more domains: ICT, Finance, Human Resources etc.
  3. Extraction of the Seniority: the document is about a senior, junior or intern.
  4. Extraction of the Work Class: the document is about an Upper Management, Lower Management,
    Office worker, or Laborer job.

2. Matching of the analyzed documents is done on 2 levels

i. On document level.

  1. Analysis of relevant concepts. Example from the vacancy: “should be familiar with a scripting language”,
    “scripting language” has been recognized as a synonym of our concept “Scripting Tools”.
  2. Extraction of one or more domains: ICT, Finance, Human Resources etc.
  3. Extraction of the Seniority: the document is about a senior, junior or intern.
  4. Extraction of the Work Class: the document is about an Upper Management, Lower Management,
    Office worker, or Laborer job.

ii. On concept level.

  1. Example: if we look for a “software engineer”, then all concepts in the resume that have “a relationship” with that job,
    will match (and will be weighted later in the ranking process – see below).
  2. Example: if we write in a vacancy that the candidate should be familiar with scripting languages, then there will be a
    match on the concept level with “JavaScript” and/or “VBscript” at the resume side.

3. Ranking of the matching results.

For each document that has successfully passed the matching test:

  1. Each extracted concept from that document will be assigned to a specific category (working experience, extra skills,
    education), which has a predefined weight. The standard weight repartition is as follows (on a total of 100% - can be
    changed upon client’s preference):
    Working Experience : 55,5 %
    Extra skills : 23,2 %
    Languages : 11,1 %
    Education : 10,2 %
  2. All the concepts assigned to the Working Experience category, for example, will get another weight inside that category.
  3. The sum of all the concept weights gives the total document ranking percentage.
  4. Important: documents in the result set having a similar matching percentage get a “random ranking within that similar percentage block”.