Contextual Search Technology: A Game Changer in Talent Acquisition

By Shreedhar Patwari: Chief Technology Officer of Spire Technologies

Introduction

Traditional keyword-based search for matching supply to demand has serious limitations. Keyword search does not provide accurate and relevant searches as it considers just the words searched and does not understand the context and semantics around the words. It does not cater to “real world” need of advanced search, which needs to consider multiple attributes and complicated conditions on these attributes. Context and semantics of the demand and supply documents is very important and provides rich source of information. This rich source of information can be further enriched to use “related information” from context cloud which is specifically built for demand and supply space. Enrichment of words and applying context and semantics are key differentiators which aids in achieving highly accurate mapping results.

Approach and Value

Talent acquisition systems have improved by leaps and bounds over last decade but there are still ongoing issues of accuracy and relevance caused by the non-uniformity of the talent data. This is a known issue in talent systems that is lacking the capacity to identify the purpose of a requirement and context around it. In this article, we highlight the pragmatic solution which would act on the semantics of the requirements where requirement is not only given a well-defined meaning, but the meaning is understood within the context it is used.

Contextual Systems

Modern talent acquisition systems need access to data from Clients and from user interactions that come through multiple channels (web, bulk upload, web crawlers, logs, 3rd party integration etc.) and across multiple interaction sessions (web sessions, bulk sessions etc.).

There are two basic types of data that these systems analyses, first is demand and second is supply which may wish to satisfy the first. There are also two more type of data. First is agent data, an agent here can be a human or a software program that matches demand with supply. Second is, the relationship data between demand and supply after the match has had happened. The primary examples of demand, supply and agent are requisition description, candidates profile and recruiter, respectively. For the simplicity, we will refer these corresponding terms interchangeably throughout the rest of the documentation.

The job of the recruiter is to match requisition with candidate. By recruiter data we mean data associated with the recruiter like the organization he belongs to, skills he is recruiting for etc. The relationship data between demand and supply gets created/appended/updated after the match has been made. The example of this data is the status of demand/supply, score, demand skills, supply skills, various statistical counts for supply/demand (total matched, total reject, interviewed and many more) and various other parameter. This data set has certain patterns in it and must be modelled in a relationship data model.

Typical Steps:

Demand and supply are both usually unstructured content. Mapping Supply to Demand involves these phases:

1. Extract structured information from Supply.
2. Extract structured information from Demand.
3. Enhance and transform structured information by considering context cloud to build Supply Context and demand Context.
4. Match the supply and demand context to each other and order the results based on the strength of match.

Conclusion

Contextual Search powered by context cloud is industry and domain neutral, which implies that the scale and possibilities of growth and application of this technology is tremendous. The application of this solution is possible anywhere there is an unstructured data in the form of demand and supply issue. Be it talent acquisition, fraud intelligence, education, dating & matrimonial solutions, compliance management and loads more. If there is unstructured data that needs to be managed, contextual search is the answer.