Eightfold's Talent Design product leverages AI to help organizations manage their talent based on skills. Here's a detailed look at how skills are extracted, used in matching and recommendations, and how equivalent skills function within Talent Management (TM) and Talent Acquisition (TA).

How Skills are Extracted from Resumes: Common Failures or Problems

Eightfold's AI-powered platform excels at extracting skills from resumes by moving beyond traditional keyword matching. It aims to resolve many common problems found in the market's approach to skill extraction and talent management.

Common Failures and Problems in Skill Extraction from Resumes in the Market:

1. Over-reliance on Keywords and Static Information:
• Many traditional resume screening tools simply compare keywords in resumes with job postings. This leads to missing qualified candidates who use different terminology for the same skill (synonyms, acronyms), or who have non-traditional backgrounds.
• Resumes are often backward-looking, focusing on past job titles and educational attainment, which are poor indicators of a person's current capabilities, potential, or learnability. They don't provide a clear picture of what a candidate is truly capable of in practice.
• "Keyword stuffing" by job seekers can artificially inflate a resume's relevance, leading to poor quality talent pools.
2. Inability to Assess Soft Skills and Context:
• Traditional software struggles to analyze soft skills like communication, teamwork, and problem-solving, which are crucial for job success.
• Resume parsers can miss context. For instance, "PM" could mean "Project Management" or "Product Management," and without intelligent contextual understanding, the system might misinterpret the skill.
3. Bias and Lack of Fairness:
• Algorithms trained on prejudiced or unrepresentative data can perpetuate and even amplify existing societal biases (e.g., gender, race, age) in hiring, leading to unfair models and missed diverse talent.
• Sticking to traditional recruiting workflows that emphasize credentials over skills can inadvertently discriminate against qualified candidates from diverse backgrounds or those with unorthodox work histories.
4. Technological Limitations and Data Inconsistencies:
• Resume parsing tools can fail to process non-standard resume formats or layouts (e.g., image-based PDFs), leading to overlooked data and the discarding of otherwise suitable applicants.
• Inconsistent skill naming (typos, variations) across resumes and within an organization's HR systems leads to fragmentation of skill data, making it hard to get an accurate count of available skills.
• Lack of explicit skill information on resumes or in HRIS/ATS systems means a limited or incomplete view of an individual's actual capabilities.
5. Difficulty in Identifying Potential and Learnability:
• Traditional methods are poor at identifying a candidate's potential to learn new skills or their transferable skills, limiting an organization's ability to hire for future needs or facilitate internal mobility.

How Eightfold Supports in Resolving These Issues:

Eightfold leverages advanced AI and a vast global dataset to overcome these common failures:

1. AI-Powered Contextual Understanding and Inference:
• Beyond Keywords: Eightfold's AI uses Natural Language Processing (NLP) and deep learning to understand the meaning of the text within a resume, rather than just matching keywords. It can interpret language that indicates a skill, even if not explicitly listed.
• Inferring Skills: The AI can infer skills a candidate likely has based on their job titles, companies, and the explicit skills of individuals with similar career experiences from its massive dataset of career trajectories, roles, and skills. This helps uncover "hidden" skills and potential.
• Contextual Ambiguity Resolution: While not encouraging abbreviations in its taxonomy, Eightfold's AI is designed to better parse context, allowing for more accurate interpretation of terms like "PM" by analyzing surrounding information in the resume.
2. Standardization and Aggregation for Data Quality:
• Skill Taxonomy and Equivalency: Eightfold maintains a standard skill taxonomy and uses "equivalent skills" to consolidate variations (e.g., "Research & Development," "Research Methods," and "Basic Research" are mapped to "Research"). This standardizes skill data across the platform, improving data accuracy and analytics.
• Automated Standardization: For unclaimed employee profiles and fresh candidate resumes, skills extracted are automatically standardized, removing duplicates and normalizing naming. This ensures "noise-free analytics."
• Search Suggestions: The platform prompts users to use standard skills in search filters, requisitions, and profiles, promoting consistent skill usage.
3. Fairness and Bias Mitigation:
• Bias-Free Screening: Eightfold's AI is designed to screen talent instantly and without bias by focusing solely on qualifications, skills, and learnability, rather than traditional proxies like education or past job titles that can carry inherent biases. This broadens the talent pool and helps identify best-fit talent irrespective of background.
• Responsible AI: Eightfold emphasizes responsible AI development, ensuring models are trained on representative data across protected categories to avoid perpetuating social injustices or biases.
4. Holistic Talent View and Potential Identification:
• Comprehensive Skill Extraction: Skills are extracted from various resume fields (job titles, education, certifications, projects, work experience, awards, patents, publications), providing a more complete picture of an individual's capabilities.
• Beyond the Static Resume: Eightfold moves beyond static resumes by understanding learnability and potential, which is crucial for internal mobility, upskilling, and matching individuals to future roles.
• External Data Integration: The AI also scours publicly available online information (e.g., social media profiles) to find additional skills, enriching the candidate's profile.

By addressing these common failures, Eightfold's approach to skill extraction from resumes provides a more accurate, comprehensive, and unbiased understanding of an individual's capabilities, enabling organizations to make more informed talent decisions.

How Skills are Used in Matching and Recommendation

Skills are central to Eightfold's matching and recommendation capabilities across its platform:

• AI-Powered Matching: Eightfold's AI uses a skills-based approach to match candidates to open positions and employees to internal opportunities. It goes beyond keyword matching by understanding skills and skill learnability, considering all qualifications to screen talent instantly and without bias. This allows for a wider talent pool and the identification of best-fit talent.
• Career Navigator: In Talent Management, the Career Navigator uses roles (defined with skills in Roles Design) to suggest career paths based on an employee's skills, title, seniority, profile data, and career interests. It also highlights skill gaps and skills present for a chosen role.
• Skill Assessments: Skills defined in Roles Design serve as the source of truth for assessing skills on an employee's profile page. All skill names and their benchmarks are derived from the associated role.
• Development Plans (Upskilling): Based on skill gaps identified through role design benchmarks, employees are encouraged to create development plans to strengthen their skills.
• Succession Planning: AI-powered matching algorithms in succession planning identify internal successors using the roles defined in Roles Design as a benchmark.
• Internal Mobility: Role calibration in Roles Design enhances internal open position calibration, helping to find the best employee matches for roles and the best jobs for employees.
• Skill Standardization on Employee Profiles: For claimed employee profiles, there is a one-time standardization of skills to remove duplicates and normalize naming. This helps ensure "noise-free analytics."
• Search Suggestions: Throughout the Talent Acquisition (TA) and Talent Management (TM) products, when users add skills to requisitions, search filters, employee profiles, or development plans, the system prompts them to use standard skills from the taxonomy, ensuring consistency.
• Market Insights: This module uses skill trends and prevalence data from a representative dataset of career trajectories to provide insights into market demands for skills and roles, helping organizations build future-ready talent frameworks.

How Equivalent Skills are Used in TM and TA

Equivalent skills play a crucial role in standardizing and aggregating skill data across Eightfold's Talent Management (TM) and Talent Acquisition (TA) modules:

• Standardization and Aggregation: In the Eightfold product ecosystem and analytics reporting, equivalent skills are consolidated under a "standard skill." For example, "Research & Development," "Research Methods," and "Basic Research" might all be considered equivalent to the standard skill "Research."
• Employee Profiles (TM):
• Display: When a resume is uploaded with an equivalent skill (e.g., "Research Methods"), the employee's profile will display the standard skill ("Research").
• Enhance Skills Functionality: For previously claimed profiles, users are offered an "Enhance Skills" option to convert equivalent skills into their standard skill counterparts, removing duplicates and normalizing naming.
• Unclaimed Profiles: For all unclaimed employee profiles, skills extracted from resumes are automatically standardized.
• Impact of Deletion: If a standard skill is deleted, any equivalent skills that were standardized to it will remain standardized on claimed profiles. If an equivalent skill is deleted, there is no automatic change to revert to the original skill name on the profile.
• Analytics (TM): Data for all equivalent skills of a standard skill is aggregated for analytics. This means that if multiple employees have different equivalent skills for the same standard skill, the prevalence count will reflect the total number of employees possessing any variant of that standard skill.
• Search Suggestions (TM & TA): When users type a skill in search fields across the platform (e.g., in positions, roles calibration, employee profiles, project marketplace, development plan creation), the system suggests both the entered text and its corresponding standard skill, even if the user typed an equivalent skill. This promotes consistency in skill usage.
• Courses: When new courses are ingested, the course's equivalent skills are picked and standardized. For instance, if a course has "PM" as a skill, it will be displayed as "Project Management" if "PM" is an equivalent skill.
• Candidate Profiles (TA): Similar to employees, the ingestion of fresh candidate resumes results in the standardization of skills.
• Data Hub: In the Data Hub, users can query "employee_skills" using both skill_name (standardized skill) and original_skill_name (the original skill from the source data). If original_skill_name is an equivalent skill, skill_name will display the standard skill. If it's neither, skill_name will be null.

Equivalent skills are also utilized in recommendation engines, where these skill groups are treated similarly within the algorithms.