Use our LinkedIn Login to download this post to PDF or save it to MyLibrary!
Everyone has been hearing about AI, some have been hearing about NLP - and everyone has an opinion, belief, or thought about AI. However, that opinion, belief, or thought about AI (and/or NLP) is fully dependent upon the voices of whom everyone has chosen to listen. We are at a stage in our rate of change of technology where we have to let go of how we learned things in the past -- and step into a new stage of keeping ourselves always available to learn, no matter what we believe and/or think.
We cannot, more than ever before, solve our problems with the same thinking we used to create them (said Albert Einstein nearly 100 years ago!). We are also going to be seeing all the analytics, management, performance, and recruitment aspects of HR becoming glued together through an entirely different, much faster means of inputting, managing, and retrieving data. This is not happening in a few years, it is already happening.
Our series provides the understanding, questioning, and ideas of AI so you can start learning how to think, learn what questions to ask, and discover for yourself what you need before you need it.
Feedback, debates, discussion, collaboration and conversation are always encouraged in the comments section below... For more information about this series, use the blue contact us button on the lower right of your screen to contact us -- or if you are reading this by our newsletter, then hit the reply button to get back to us!
Text analysis is possibly the most familiar tool known HR.
Almost everyone in HR/ recruitment is hands on with Boolean keyword searches for identifying good resumes and of course the oft humorous results and typical frustrations associated with it.
Natural Language processing (NLP) basically takes text analysis to a much higher level of detail, granularity and accuracy.
Acute insights from NLP were a technological constraint in the past but there have been major strides of late.
This has primarily been aided by the development of distributed computing (cloud) and intense research in NLP applications by academic and professional bodies around the world.
R&D in NLP comes within the ambit of applied statistics research
Now, the very crux of people function lies in effective analysis of people communication and goes without saying natural language is the most prevalent medium of human communication
However, core technology apart, the scope and magnitude of applied NLP in people function has to be spearheaded by operational HR alone.
It is self evident, most HR business processes generate mammoth volumes of natural language (unstructured data), business processes as in
- employee feedback’s
- legal cases
- counseling and
- many others
Additionally, legacy HR processes and forms can be conveniently re-engineered to accumulate ever increasing volumes of natural language data, for example via an active policy of audio recording & transcribing or even a slight redesign of various HR processes forms/surveys/applications.
What are the key benefits for HR with reference to NLP?
Benefits are multifarious, corresponding to varying levels of engagement and investment.
It starts from
- generic text analytic s (sentiment analysis)
- advanced insights (via computational linguistics models), and even
- potential semi-automation
Once implemented, such services are typically delivered via
- APIs (cloud)
- remote ODBC connectivity or
- standalone client based systems
Google and Microsoft are prime big player examples in the API NLP space.
However, advanced insight systems (specific to HR)
typically need a
corroborative development program ME
in conjunction with operational HR
How do insights from NLP analysis impact HR?
Typically HR specific NLP analysis, with varying and often progressive levels of insights act as decision supports (DSS), enabling greater accuracy and speed to key HR business processes, progressively improving HR metrics.
They can also reduce human bias in decision making
Application/resume scoring and survey analysis are key examples — but the potential for applications is across the HR spectrum.
Very often NLP systems act as “first or second level filtering” or “hypothesis proofs” to corroborate human decisions in HR.
Are Voices to Future Irrelevancies Valid?
With some much buzz around NLP in HR, does it imply that structured (non text based) data may loose relevance?
Not at all, on the contrary structured and unstructured data perfectly synergize to improve quality of insights for HR
For instance, key traditional areas for HR data modeling have been attrition, absenteeism, carrier paths, compensation/ benefits, etc.
For such models, the insights gained through NLP can be modified to fit in as explanatory variable thereby improving model accuracy.
Yes, HR processes forms like employee survey, feedback, and assessments have traditionally consisted of structured data points (check boxes, radio buttons, drop-downs. slide boxes, etc).
The advances in NLP technology enable HR to further incorporate and capitalize from short free flow natural language synopsis field in these forms, vastly improving quality of insights
In essence both in conjunction improve metrics of HR processes and one isn’t a replacement for the other.
What misinformation exists with regards to the adoption of NLP in HR Processes?
There is apparently some disconcert, possibly due to a few misunderstandings — NLP, cognitive sciences, automation in HR and the resulting job losses are the more popular disconcert voices.
NLP systems are not — and cannot — be a human HR replacement, on the contrary these systems further empower HR personnel within their organization.
The complexities of human language, communication and dynamic decision making required by HR in the real world imply that total automation is impractical and can be downright counterproductive.
Machines find it extremely complex to comprehend the finer nuances and ambiguities of human language such as sarcasm, ambivalence, deformed compliments, passive aggression, regional norms.
An interesting and somewhat parallel comparison is in the case of autopilots and fly by wire systems — they have been with us for more than two decades yet never replaced humans in cockpits, even though in flight simulation tests they routinely outperform human pilots.
Drones — theoretically pilot-less — are always controlled by a human pilot
Therefore concept of full automation is completely misplaced.
Job losses are opportunities for
skill restructuring / retraining / realignment programmes
and may be progressively required.
It is imperative to take the initiative here.
What are the other bottle necks in adopting NLP to HR processes?
There aren’t many vendors who are exclusively focused on advanced NLP to HR processes as yet — most NLP vendors are text analytic generalists, they may not have in-depth awareness aware of HR specific problems.
Organization View is a good example of a dedicated operator in this space, there are a few more (editor’s note: see our Enterprise Software Options Series for WhoKnows, Inc. who is a very strong leader in this accomplishment).
Other key bottlenecks are HR data security, data accessibility, quality, APIi integration.
The engagement/ collaboration programmes between HR technologist and Operational HR also has scope for improvement in this area.
Interestingly, large strides have been made in recent times pertaining to the application of NLP to clinical, legal, and entertainment businesses (script writing, for instance) and the time is ripe for its adoption to HR.
NLP Relevancies For and To HR
What are the identified technologies/approaches in NLP that are relevant to HR and in which particular areas do they improve HR functionality?
Ideally, operational HR should take the lead and collaborate to identity relevant application areas within their own organizations.
The leverage of NLP in HR is likely to depend upon data availability, security, integration, company policy, and / or specific business requirements.
Broadly there are 3 aspects to applying NLP to HR…
- Types of Generic NLP insights ( relevant to the HR application context) can be enumerated as…
Sentiment Analysis of HR documents
Deep Information Extraction from HR documents
Classification/ ranking of HR documents as per business specifications
Automated Summation of HR documents ( topic discovery )
Establishing HR Hypothesis and process improvement ( a part prescriptive analytics )
- Application areas of NLP ( within the HR application context )
Application/ Resume scoring
360 degree feedback analysis
Multiple HR surveys analysis
NLP based insights on appraisals
Social media content analysis of employees
Insights on documented Legal cases / suits
Design and insights about Employee Counseling
NLP on virtually any unstructured data within the ambit of HR, including transcribed data
- Broad overview of various NLP methodologies employed by NLP vendors ( within the HR application context )
Also referred to as POS tagging.
Statistical surgery on it offers insights from various levels of granularity starting from basic text classification, sentiment analysis to deep information extraction and topic modeling / automated summation.
Some of the popular information extraction / topic discovery approaches are Conditional Random Fields, Hidden Markov Models and LDA.
The HR familiar basic Boolean keyword searches to identify good resumes, is a very good example of symbolic tagging
However, today NLP models like nested, iterative, and conditional “regular expressions” can fine tune symbolic tag searches to the deepest possible levels of granularity.
Searching for a needle in a haystack, no problem!/p>
A combination approach of statistical and symbolic tagging is often referred to as a “conditional rules model” within the NLP context
Importantly tailored combinations of “conditional rules models” are typically developed via integrated cohort analysis in collaboration with HR.
This may be also useful to establish evidence based HR Hypothesis and effectively push forward major HR initiatives to the Company leadership.
post continues after these free offers
Today's Featured Free Offer
Communication is more than just a monthly newsletter.
A Modern Communicator's Guide to Corporate Communications
This 6-page guide is full of tips and tricks to help you create compelling stories and enable meaningful conversations inside your organization with the help of a modern intranet solution.
Offered Free by: Igloo Software
post continues from above
Could you give a production example of NLP in a key HR process?
The basic approach of NLP remains more or less the same across all types of HR unstructured data, however for the sake of familiarity lets take the example of resume scoring on an large unstructured data-set.
Here is where NLP can help…
- Automatically classify and rank resumes according to their core skills, experience or any other priory ( for example desirable skills and professional experience )
- Automatically classify resumes according to their format styles ( chronological, reverse chronological, hybrid, skills based, qualification based functional based format
- Identifying basic sections of a resume ( topic model based on the prirority given by HR)
- Potentially identify gaps in professional/ academic records in resumes
- Identify potential fraud/ incorrect information and anomalies in resumes
- Deep information extraction from resumes ( for instance combination of professional skills / education + university rankings + professional experiences + environment and context + international assignments / location specific + awards / recognitions + recommendations / professional network ) via compound “conditional rules models”
Apart from resume / application scoring “Conditional rules models” can also help identify complex human language expressions like sarcasm, ambivalence, deformed compliments, passive aggression, this might be important for HR surveys, feedback’s, forums, social media data etc.
However the degrees of accuracy are subjective here (and it’s an ongoing research area).
In systems where applications / resumes have semi structured data points ( for example applications received through an online XML form ), NLP can act in conjunction with the structured data points ( SQL) for improving the quality and accuracy of classifications and inferences.
Once developed , semi automation can also be applied to NLP models to enable
- Periodic and automated evaluation of data set via cron jobs and database procedures/ triggers/functions
- Automated scoring and classification of data sets via above
- Sending automated email to shortlisted candidates ( for example a test set or interview call ) or sending a consolidated or specific reports to the Human Resources team
It would also help them have informed discussions with NLP vendors with respect to customizations required and be able to discern between their services and methodologies.
What are the typical types of services offered by NLP vendors ?
NLP vendors typically offer a combination of services mentioned above.
High end insights ( summation, topic modeling, conditional rules models ) may be offered under a beta testing/ release model.
Innovative marketing and promotion methodologies might give the impression a variety of computational approaches in NLP among different vendors to the end user, while fundamentally being the same.
Hence it is important for operational HR to be able to have a good overview and be able discern NLP applications relevant to their specific business requirements and constraints.
NLP application service delivery might be through API web services / ODBC integration (cloud model) or standalone implementations on clients ( windows based installations ) or even static periodic reporting system (for example, process improvement via a six sigma framework).
How should HR maximize leverage from NLP ?
A productive implementation of NLP in HR processes will ideally be a cohort, collaborative exercise between the NLP vendor and operational HR.
Once the framework around NLP technology relevant to HR is understood, it enables the formation of a structured engagement program me and a stepwise integration with key HR business processes/data.
This may involve some HR process redesign and data integration, protocols.
The implementation should ideally be within the ambit of a well structured implementation program me (like six sigma or CMMI process improvement program me).
It might be required to integrate with existing cloud based HRIS systems (like Sucessfactors, Kenexa, Deollite) via API /ODBC.
What does the future hold in terms of the relevance of NLP in HR?
HR is the prime candidate for adoption of NLP based technologies due to its very foundation based in “people centric function & people communication”.
Most HR business processes generate vast amount of natural language data.
The onus remains on operational HR to effectively tap this technology via corroboration with an HR technologist and canvass for its benefits to their leadership.
This also presents a key opportunity in realizing the goal of HR towards becoming a profit center, it will also enable HR to have greater intelligence and leverage within their own organization .
Early adopters and pioneers, please do contact us either through our LinkedIn profiles (links in our bios below) or use the contact us button to the lower right of your screen!
Editor’s note: Learn how text analytics can help your organization gain significant, measurable benefits from textual data from our free download Mastering New Challenges in Text Analytics
Our Social Media Presence
Where to Follow Us!
Soumyasanto Sen and Raja Sengupta
Advisor/Consultant/Investor in HR Technology currently focusing on AI driven People Analytics, Digital Strategies, and Digital HR Transformation. Certified ITIL and PRINCE 2 Project Manager.
Blogs on: SAP Community, Data Science Central, HRTech Conscience, SAP Startup Focus, Analytics in HR, HRTech Weekly, HR Gazette, WISP.
Speaker at: SAP Info Day-Walldorf, SAP Inside Track-Frankfurt, SAP Inside Track-Munich, SAP Inside Track- Walldorf, Entrepreneur Meetup- Frankfurt
Leveraging entire data science experience of 17 years for the HR Domain. Leading innovation and change manager for global HR.
17 years of R&D and consulting experience in the entire spectrum of applied statistics, analytics, six sigma, computer programming and NLP with focus on on the HR domain. HR technologist , thinker, innovator, developer and writer.
One of the most internationally networked HR analytics consultants from India. More than 500 key global HR connections including top industry contacts. Collaborated and shared data science insights with global HR analytics leaders via remote consulting.
Published HR R&D whitepapers in Europe, India and Canada . Current focus on the application of deep NLP learning to HR process data.
Integration of HR metrics with a six sigma framework for project management and process improvement. Currently in R&D stage of a product development with focus on application of NLP to HR.
Latest posts by Soumyasanto Sen and Raja Sengupta (see all)
- What You Need to Know About AI and NLP When It Comes to HR - Mon, 27-Mar-2017