Skip to main content
Get Started
Avature

The Brains Behind Avature AI

Our in-house team of over 50 AI experts is actively shaping the future of artificial intelligence by conducting novel research while continuously enhancing Avature’s capabilities.

Machine Learning (ML) Team

The ML team sits at the heart of our AI innovation and is in charge of optimizing our models through cutting-edge research and development.

Natural Language Processing (NLP) Team

The NLP team is responsible for improving features such as our matching engine, skills framework and resume parser.

Chatbots Team

The Chatbots team is focused on developing and maintaining our conversational AI chatbot and constructing a database to continue improving our language processing and generation capabilities.

AI For Talent Management: Our Team’s Research

In addition to partnering with leading academic institutions to tackle HR’s most pressing challenges, we actively contribute to the AI community by publishing original research in peer-reviewed journals and presenting our advancements at scientific conferences.

2024

MELO: An Evaluation Benchmark for Multilingual Entity Linking of Occupations

MELO offers 48 datasets to evaluate entity linking of occupations across 21 languages using AI models. This benchmark helps standardize multilingual HR data normalization, supporting organizations in streamlining processes and improving global hiring.

Go to PDF File
2024

Intent Classification Methods for Human Resources Chatbots

This paper analyzes intent classification techniques applied to HR chatbots. Besides exploring supervised and unsupervised learning methods, the study proposes a two-stage retrieval pipeline that increases performance and flexibility by allowing the addition of new intents without retraining.

Go to PDF File
2024

Inductive Graph Neural Network for Job-Skill Framework Analysis

This study showcases the power of combining text encoding and graph neural networks to analyze job-skill links, enabling more accurate job and skill recommendations. It excels with structured and unstructured data, even when dealing with unseen skills.

Go to PDF File
2024

Combined Unsupervised and Contrastive Learning for Multilingual Job Recommendation

This study explores a multilingual job recommendation model that improves ranking accuracy across 11 languages. Its two-stage learning approach supports cross-lingual recommendations and outperforms existing methods in aligning job titles globally.

Go to PDF File
2023

Normalization of Education Information in Digitalized Recruitment Processes

Extracting information from a resume is usually a two-stage process. Our ML team proposes using a neural network architecture to tackle both stages simultaneously. The model improves system efficiency and resume parsing accuracy in 7 languages.

Go to PDF File
2023

Resume Parsing as Hierarchical Sequence Labeling: An Empirical Study

In this paper, our team normalizes education information extracted from resumes by transforming it into level/field of study. They also define a new taxonomy for fields of study and show how the process can be applied to candidate-job matching.

Go to PDF File
2022

Learning Job Titles Similarity from Noisy Skill Labels

Other ML models measure the semantic similarity between job titles using supervised learning techniques trained on vast amounts of manually labeled data. We propose an unsupervised representation learning method that is as effective and less costly.

Go to PDF File

Driving AI Standards With TalentCLEF

Avature has spearheaded the creation of TalentCLEF, a pioneering global initiative hosted by the CLEF organization to advance Natural Language Processing (NLP) in HR. As the organizer, we are shaping benchmarks that enhance AI fairness, multilingual adaptability, and real-world relevance in hiring.

This endeavor reflects Avature’s commitment to shaping the future of HR technology, ensuring AI supports – rather than replaces – human expertise.

Learn More

Additional Resources

A Deep Dive Into Avature’s Job Titles Similarity Model

Our Machine Learning research team has developed an innovative approach to determine the semantic similarity between job titles, and we’ve gathered the most noteworthy points of that work in this blog.

View

Making Sense of Skills: Neural Network Models for Skills Semantics

Avature’s AI expert, Rabih Zbib, gives us an in-depth look on how skills semantics and embeddings work with the help of AI and machine learning.

View