Knowledge Graph

Introduction

To understand and use the relations between different entities such as job resumes, companies, etc. we have created a Knowledge Graph. The KG contains entities like jobs, resumes, skills, industries, companies, etc, and their edges which depict their relations with other such entities. The KG is enriched from Turbohire’s in-house data as well as publicly available data. The KG contains almost 3.5M nodes and 50M edges.

• Different types of nodes with different properties are all connected with each other by different connections.
• 2 different nodes can be directly or indirectly connected with other nodes and edges in between them.
• It is very easy to find similarities between 2 nodes on the basis of their properties and common nodes between them.
• Eg: How much companies are similar is based on their employee count, age, industry, roles they hire, etc.
• 1.6M resumes, 1.2M companies, 100K skills, 30k jobs, and 10k industries are present in the Knowledge Graph.

Problem Statement

1. Screening gap for a recruiter understanding every aspect of a Resume
2. Augment your understanding of a job using our dynamic KG data which shows a more clear picture of the current scenario

A typical resume contains skills, companies, experience, education data, candidate roles, locations, etc. It is not possible for a recruiter to know about all these aspects of the resume during the hiring process. Minor differences between the actual value and the understanding of the recruiter for these aspects aggregate and make it very difficult to evaluate a candidate correctly.

Also, it is important to always be up-to-date with this knowledge because of the ever-changing job market. A skill that was very important for a role might not be as important right now. So, even when understanding is clear, it is very important to know how a skill fairs in the current hiring scenario.

Usecases

• Skill Suggestion: Suggests skills based on which skills were required in the past for a specific job title
• Industry Suggestion: Suggests similar industries based on the industry of the recruiter’s company
• Company Overview Card: Returns multiple properties corresponding to a company such as an overview, ratings, size, age, etc helpful for understanding a company

Solution

With all nodes having multiple properties as well as connections with other nodes, the knowledge graph has enough information to help the recruiter make better choices when selecting a resume.

Candidates that have skills, industries, and other nodes similar to the requirements of the job will have an edge over other candidates. With the help of properties and connections that these nodes have, we can show which nodes are similar to each other. Since there are many types of nodes in the graph, we can find similarities between nodes of different types.

Eg: A recruiter can know resumes from a specific company have a high density of a specific skill. This would significantly affect the results when comparing multiple resumes having different past companies.

Similarly, by getting to know the density of edges emerging from a node, we can get an idea of how important that entity is in the current job market. The data in the graph is automatically increasing on a daily basis so the results are always up-to-date.

In skill suggestions, we suggest the most common and useful skills that were required by similar job titles and skills present in candidates with similar roles. More important skills would be among the top suggestions.

In Industry Suggestions, we suggest the industries that have candidates that should have skill sets good for the job based on the recruiting company and job title.

Company Overview card shows multiple details of a company that might be present in the past experiences of the resume. The card shows an overview, employee count, founded date, and multiple ratings based on work experience and culture, salary benefits, etc.