ML Skill Inference System Using AutoML and LDA Topic Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current skill frameworks in large companies are often incomplete and outdated, failing to accurately identify the presence or absence of skills and proficiency levels among employees, despite their acknowledged value for job roles.
Innovation Solution
A system comprising a skills data store, employee action data store, and software modules that preprocess employee action data and skill descriptions using machine learning techniques, including feature preprocessing, LDA topic vectors, TF/IDF Word2Vec similarity scoring, and AutoML to train models for predicting the presence and proficiency of skills.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional skill frameworks are used to identify employee skills, then skill identification is acknowledged as valuable for job roles, but the frameworks become incomplete and outdated over time
Solution Approach 1:
The system enables automated self-updating of skill frameworks through machine learning models that continuously learn from employee action data. The classification models automatically retrain and adapt to new skills and behaviors without manual intervention, allowing the skill framework to serve itself and remain current with evolving employee capabilities
Solution Approach 2:
The system implements continuous feedback loops where employee action data is constantly fed into the classification models, which then update skill predictions. This feedback mechanism ensures the skill framework evolves based on actual employee behaviors and achievements, maintaining both accuracy and timeliness
2Adaptability or versatility
If comprehensive skill frameworks with hundreds of specific skills are created, then skill coverage is improved, but the complexity of tracking and identifying skills increases
Solution Approach 1:
The patent replaces manual, mechanical processes of skill tracking with automated machine learning classification models. These models automatically analyze employee action data, perform feature preprocessing, generate topic vectors, and predict skill presence and proficiency levels without human intervention, significantly reducing system complexity while maintaining comprehensive skill coverage
Solution Approach 2:
The classification models serve multiple functions simultaneously: they identify skill presence, determine proficiency levels, track skill evolution over time, and adapt to new skills. This multi-functionality allows a single unified system to handle comprehensive skill frameworks without proportionally increasing complexity
3Measurement precision
If machine learning models are trained on organization-specific data, then model accuracy for that organization is improved, but overfitting occurs and reduces generalizability
Solution Approach 1:
The system implements dynamic model training where classification models are continuously retrained with new organization-specific data while maintaining their ability to generalize. The models adapt to each organization's unique characteristics through ongoing training on fresh action data, achieving both high accuracy and sustained generalizability through continuous adaptation rather than static over-fitting
Data Source
AI summary
A system comprising: a skills data store; an employee action data store; at least one hardware processor; and one or more software modules that are configured to, when executed by the at least one hardware processor, retrieve skills data and employee action data from the skills data store and employee action data store, train a classification model, wherein training a classification model comprises performing feature preprocessing, generating an LDA topic vector and TF/IDF Word2Vec similarity scoring, and use AutoML to train ML models, and infer employee skills and levels based on the classification model and employee action data.