Before the establishment of the model, a detailed analysis is conducted based on the collection of CVs, job postings, and the assignment of which CV goes to which job posting. The parameters and ratios identified to help pinpoint the right candidate are as follows:
- Educational Background (Level of education, university information, field of study and the knowledge acquired in that field, training and courses attended, thesis and research projects, and the skills developed through these projects, etc.) - 20% weightage.
- Professional Experience (Industry experience, job position experience, roles and responsibilities, the relationship between job description and responsibilities, years of experience, etc.) - 30% weightage.
- Technical and Job Skills (Technical specifics related to the job sector: programming languages like SQL, Python; software such as Looker, Tableau; capabilities like pipeline creation, customer management, communication skills, etc.) - 30% weightage.
- Job Responsibilities and Achievements (Teamwork, problem-solving and analytical thinking abilities, adaptability to company culture, project management, leadership qualities, accomplishments, KPI results, and industry-specific achievements, etc.) - 20% weightage.
Based on the collected data and parameters, language model training was conducted. This new model is a self-learning dynamic structure that determines parameter weights itself, but it is trained according to the ratios mentioned above. The model is also sensitive to adding new weights or weighting the items written in the job posting internally, to meet this requirement, new weights need to be added for each job posting feature conveyed to the job posting details entered into the system.
Following this parameter and weighted evaluation training, our HR recruitment language model analyzes CVs and job postings and produces a compatibility percentage. This indicates that our model is continuously learning and dynamically structured, able to evaluate each candidate's unique attributes in a more sophisticated manner. A comprehensive analysis of CVs and job postings is performed semantically rather than based on rule-based structures, providing a detailed report for each CV that includes:
- CV and job posting match score.
- Strong points of the candidate concerning the job posting.
- Weak points of the candidate concerning the job posting.
Our value proposition is being the first in the HR sector to perform CV and job posting reviews at this semantic dimension.