Text – and the challenges it poses in recruitment technology
- Resumes and Job Descriptions are two key documents that are used in recruitment technology. Both are highly textual in nature.
- The texts in these, particularly when it comes to skills, pose a lot of challenges in matching people to jobs.
- Here we will analyze the various issues caused by texts relating to skills
In our earlier article – Standardization of Data on Skills in the HR Space – we touched upon the importance of standardization. In this article we dive a little deeper into the problems related to textual data on skills in the talentscape.
The analysis in this article focuses on the Recruitment function since it is easy to relate to, and also, it a function where the implications of the form of skills data are huge. However it is important to note that the same issues also impact other talent functions including development planning, resource planning and deployment, knowledge sharing and skills analytics.
In Recruitment, there are two major and widespread problems:
Improper matching between jobs and people
The wrong people are frequently picked (sourced, screened and shortlisted) for a job.
Often a person who would actually be a good fit for a job is missed out in the sourcing, screening and shortlisting process
To a large extent these problems are the result of the basic constructs that everyone today uses to express skills, i.e., the textual format of resumes and job descriptions that are used to express the skills possessed by individuals or the skills required for jobs respectively.
Of course, there are other issues that exacerbate the problems in recruitment, but these will be covered in detail in subsequent posts.
There are different problems associated with the textual nature of resumes and job descriptions that are worth analyzing.
People often express the same thing in different ways. For instance, take ReactJS – different people may refer to this skill in different ways – ReactJS, React.JS, React JS. This may seem a trivial difference, but it has profound implications.
Think of a recruiter searching for the phrase “ReactJS” as a requirement for a post he wishes to recruit for. Many resumes where the same skill has been written as React.JS or React JS would be missed, some of whom may be good candidates for the role.
This is a widespread problem, and given the sheer number of skills out there which different people may express in different ways, the scale of the issue is immense.
Added to this is the very basic, but no less crucial, issue of usage of different standards of the English language – US and UK – and the different spellings of words associated with them.
The same skill may be referred to in different words or phrases with the same meaning. To take the example of Digital Marketing, the same is referred to as Internet Marketing, Web Marketing, Online Marketing and so on by different people, but meaning the same thing.
Again, this is a major problem because searching for one of the synonymous words or phrases will mean that the others – which could also be potential opportunities or good fits for the requirement – could be left out.
The usage of acronyms – especially for commonly used phrases – is a natural, almost instinctive process. A person may say USA, or even US, for the United States of America.
However, when different documents – for instance, resumes and job descriptions – do not consistently use acronyms, it becomes difficult to find and match these skills. This is another widespread issue faced in recruitment technology.
For instance, a set of resumes may use the acronym “DBA”, while the recruiter may search for the phrase “Database Administrator”, resulting in many relevant resumes being missed out.
Words, in conjunction with other words, can take on different meanings. For example, take “Oracle” – a person could be involved in Oracle Development or Oracle Administration or be a Consultant in Oracle Applications (which itself could be referred by some as Apps).
Or to look at another case, take example of “Boilers” – a person could be involved in Designing or Engineering of Boilers or Maintenance of Boilers or Operation of Boilers.
The skills involved are different in each of these cases. The probability is high that human beings, as well as parsers, would miss out on this contextual reference – for example, searching only for “Oracle” or “Boilers” can lead to a mismatch of jobs and people.
Same word or phrase understood differently
Each person has notions of what a particular phrase means based on their experience. And any two people could have a different interpretation of a given phrase based on different experiences.
For example, take two people working in two different companies. For one of them, a Project Manager’s responsibilities include delivery management, client management and also Profit & Loss (P&L). Not so for the other, in whose company, a Project Manager’s responsibilities are only delivery management and related activities. It is quite natural that when either one of these people articulates a requirement for filling a Project Manager position, they refer to two dissimilar sets of activities.
Many inefficiencies in recruitment technology occur due to such differences in understanding of phrases.
All of these issues are related to the textual nature of legacy formats for articulating skills – resumes and job descriptions. In the talentscape we need to tackle this problem in order to make the process of matching skills of people to that of jobs. Maybe we need to consider alternatives to resumes and job descriptions itself. It’s Your Skills attempts to do this having created a Skills Ontology and a Skills Profiler that can be universally used for creating standard Skills Profile for mapping of skills of jobs and people alike.
In the next post we will explore the issues of information flow in the recruitment process and analyze the problems they cause.