1.

​ Introduction Slide-

Today's career landscape presents unprecedented challenges. In an increasingly dynamic and

competitive job market, the career choices we make profoundly impact not only our professional
success but also our personal fulfillment and psychological well-being. Effective guidance is
more essential than ever to help individuals align their skills and interests with real opportunities.

However, traditional counseling methods face significant limitations. Conventional approaches

typically rely on standardized aptitude tests and generic advice, which fail to provide truly
personalized guidance. They rarely incorporate real-time labor market data or account for
emotional and motivational factors—often leading to suboptimal career outcomes.

Adding to this is the growing complexity in decision-making. Students today encounter an

overwhelming array of career options within rapidly evolving industries. Combined with
inadequate guidance, this results in decision fatigue, uncertainty, and misaligned career paths
that diminish long-term satisfaction.

2.​ Objectives Slide-

Our study is driven by three key objectives.

First, we aim to review and evaluate existing AI and machine learning-based models that are
currently being used in career recommendation and educational guidance systems.

Second, we focus on identifying major research gaps—specifically the limited personalization in

current systems, concerns around ethical fairness, and the poor alignment of recommendations
with today’s rapidly evolving labor market.

And third, we propose a unified, Ikigai-based conceptual framework. This framework brings
together career aptitude testing, psychometric profiling, and real-time market analytics—to guide
the development of more adaptive and meaningful career guidance systems in the future.

3.​ Overview of Career Recommender Techniques-

This slide highlights the five key technique categories used in modern AI-driven career
recommendation systems.

Machine learning algorithms form the foundation, with Decision Trees (DT), Random Forest
(RF), Support Vector Machine (SVM), Naïve Bayes (NB), and K-Nearest Neighbors (KNN)
effectively classifying career paths based on student attributes.

Deep learning approaches have advanced the field through specialized neural networks like
TAPJFNN (Task and Preference Joint Factor Neural Network), which excel at capturing complex
relationships between user profiles and job requirements.
Graph-based models represent the cutting edge, particularly bilateral heterogeneous graphs
with attention mechanisms that consider both candidate and employer perspectives
simultaneously.

Hybrid recommender systems combine multiple techniques, integrating collaborative filtering

(CF) with content-based approaches (CBF) and knowledge-based methods (KBRS), often
achieving accuracy rates above 88%.

Finally, reinforcement learning (RL) and fuzzy logic approaches better handle uncertainty and
changing market dynamics, making recommendations more adaptable to real-world conditions.​
​

Comparison table
This slide provides a synthesized overview of recent advancements in AI- and ML-based career
guidance systems, highlighting key methodologies, data sources, and performance outcomes.

Across the literature, a diverse range of machine learning and deep learning techniques have
been employed. Traditional models like Support Vector Classifiers, Random Forests, and Naive
Bayes have demonstrated solid accuracy, often exceeding 85%. Meanwhile, deep learning
models—including GRUs, CNNs, and attention-based architectures—have enabled more
nuanced analysis of sequential and unstructured data, such as resumes and job descriptions.

Hybrid models are particularly prominent, combining content-based and collaborative filtering
with machine learning algorithms to enhance personalization and adaptability. These systems
often integrate academic records, user preferences, and personality traits to improve
recommendation relevance. Some models further incorporate user feedback loops or
reinforcement learning to align outcomes with evolving career goals and labor market demands.

Graph-based approaches and metapath attention mechanisms have proven effective on

large-scale datasets—sometimes involving over 170,000 samples—capturing complex relational
structures among users, jobs, and skills. These techniques excel in modeling competition and
reciprocity within job markets.

Several studies also tackled persistent challenges such as the cold-start problem and data
imbalance. Solutions ranged from convolutional neural networks paired with NLP techniques to
ensemble strategies using AdaBoost and genetic algorithms, with F1 scores reaching as high as
98.3%.

User satisfaction and practical impact were frequently assessed. Some systems reported
satisfaction rates above 90% and demonstrated tangible outcomes like increased salary offers
and improved skill diversity among candidates. Others validated effectiveness through pre- and
post-intervention testing or by measuring alignment between recommendations and student
interests.
Overall, the literature indicates that hybrid and deep learning approaches—particularly those
incorporating contextual and personal data—tend to yield the most effective and user-aligned
results. These systems not only improve recommendation accuracy but also support broader
educational and career development outcomes.

Research Gaps

Data Limitations​
“Career recommendation systems often face the Cold-Start Problem, struggling with new
users or careers due to limited data. Data Sparsity makes it hard to build accurate models, and
the Grey-Sheep Problem affects users with unique preferences who don’t match common
patterns.”

Personalization Gaps​
“There's a clear Lack of Deep Personalization—most systems overlook user values, local
context, and goals. Feature Overload also adds complexity without improving results.”

Slide 3: Contextual Awareness​

“Many tools show Limited Context Awareness, failing to adapt to market trends or changing
user needs. They also use Outdated Occupational Taxonomies, which don’t support new or
hybrid careers.”

Slide 4: Bias and Inclusivity​

“Cultural & Multilingual Gaps limit accessibility for diverse users. Plus, Algorithmic Bias can
reinforce harmful stereotypes, like gender-based career suggestions.”

Slide 5: Algorithmic Issues​

“Finally, Overfitting & Poor Generalization mean models often don’t perform well across
different regions or user groups, limiting scalability and fairness.”

Ikigai-Based Career Recommendation Model – Concept and Benefits

The proposed model is a holistic, AI-driven system inspired by the Ikigai concept, which
highlights the intersection of what an individual loves, what they are good at, what they can be
paid for, and what the world needs.

This approach goes beyond conventional methods by aligning personal strengths and passions
with real-world opportunities. It emphasizes not just capability, but purpose and fulfillment.
By combining data-driven assessments with qualitative insights, the model supports better
career satisfaction, improved job performance, and overall well-being. It’s designed for both
students and professionals looking to make informed and future-ready career decisions.

The ultimate goal is to provide personalized, relevant, and practical recommendations that
adapt to evolving market trends while staying true to the individual’s identity and potential.

Slide 8: Structure of the Model – The Three Integrated Modules

The model operates through three core modules, each targeting a key dimension of career fit:

First, the Career Aptitude Test assesses cognitive abilities like analytical thinking, creativity,
and technical skills. Responses are mapped to career clusters using machine learning,
producing a ranked list of career options based on individual strengths.

Second, Psychometric Analysis evaluates personality traits and work preferences through
tools like MBTI and Holland Codes. AI-based pattern matching links these traits to suitable job
roles and environments, ensuring a good fit beyond just skills.

Third, the Real-Time Job Market Trends module analyzes current labor data from platforms
like LinkedIn and Glassdoor. It identifies in-demand careers, required qualifications, and salary
insights to ensure recommendations are practical and economically viable.

All three modules are integrated through a weighted scoring algorithm that filters and ranks
results. The outcome is a refined list of three career recommendations tailored to the individual’s
abilities, interests, and market demand.​
​
​
​
Slide 9: Trends & Directions

“As we look ahead, several key trends are shaping the future of AI in career
guidance.”

1. Emphasis on Explainable AI

“First, there's a growing focus on explainable AI. One of the biggest challenges in
AI-based recommendations has been the lack of transparency—users often don’t
understand why a certain career is suggested. Explainable AI aims to solve this by
making the reasoning behind recommendations clear and easy to understand,
which builds trust and encourages adoption.”

2. Integration of Real-Time Labor Data

 “Next, we're seeing the integration of real-time labor market data. By pulling
information from job platforms like LinkedIn and Glassdoor, AI systems can now
recommend careers based on current market demand, salary trends, and even
geographic needs. This ensures that recommendations are not just a good fit—but
also relevant and future-proof.”

3. Inclusion of Psychological and Emotional Well-being

“Another important trend is the inclusion of psychological and emotional well-being.

Modern systems don’t just focus on skills—they also consider personality, emotional
traits, and interests. This holistic approach helps align users with careers that not
only match their abilities but also their passions, promoting long-term job
satisfaction and mental wellness.”

4. Focus on Fairness and Ethical AI

“Finally, there’s a major push toward fairness and ethical AI. Historically, some
systems have reflected existing societal biases—like gender stereotypes in job
roles. Today, there’s a conscious effort to eliminate that bias and ensure that career
recommendations are inclusive, unbiased, and equitable for all users.”

Slide Conclusion
In conclusion, this review underscores the significant impact of AI and ML in reshaping career
counseling, with existing literature demonstrating promising results through data-driven and
personalized recommendations. However, challenges such as the cold-start and grey-sheep
problems, data sparsity, and algorithmic bias persist. This review lays the foundation for
developing ethical and holistic AI-driven solutions, exemplified by the proposed Ikigai-inspired
model that bridges quantitative data from aptitude and market analysis with qualitative aspects
of personality and aspirations to provide more meaningful guidance. Our future research would
be focused on achieving scalability, implementing fairness-aware AI, and ensuring the inclusion
of new and blended professions to fully realize the potential of AI/ML in creating a more
balanced and effective career landscape.