Constraint-Aware Conversational Job Recommendation in Code-Mixed Low-Resource Settings

2026-10-05

AI Summary

This study addresses the issue in low-resource code-mixed conversational job recommendation where strict constraint filtering erroneously discards suitable positions. To mitigate this, we propose W-SCAR, a weighted soft-constraint ranking framework that integrates BM25 with multilingual dense retrieval and employs TOPSIS-based multi-criteria decision-making to softly rank candidates by lexical, semantic, and utility attributes, thereby avoiding destructive pruning. Furthermore, we construct JobCCC, the first Bengali code-mixed (Banglish) conversational recommendation benchmark, annotated with LLM assistance. Experimental results demonstrate that W-SCAR achieves Hit@10 scores of 37.37% and 38.43% in English and Banglish settings, respectively, significantly outperforming strict-filtering baselines.

Abstract

Conversational job recommendation requires jointly modeling semantic relevance, user preferences, eligibility requirements, and the noisy language used in real-world career discussions. These challenges are especially pronounced in low-resource, code-mixed settings, where strict constraint matching can incorrectly eliminate otherwise suitable jobs. We introduce JobCCC, a conversational job recommendation benchmark for Bangladesh comprising 22,410 structured job postings and 988 multi-turn career-advice dialogues derived from regional Reddit communities. Each dialogue is annotated with evolving seeker preferences and linked to a ground-truth job, and is evaluated in semantically equivalent English and Romanized Bangla--English variants. We compare sparse BM25 retrieval, multilingual dense retrieval, and their hard-constraint-filtered counterparts against Weighted Soft-Constraint-Aware Ranking (W-SCAR), our multi-criteria ranking framework that combines lexical relevance, semantic relevance, and graded utilities for experience, location, education, and salary using the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). Experiments reveal that strict filtering consistently degrades retrieval because incomplete extraction and brittle attribute matching irreversibly remove relevant jobs. W-SCAR avoids destructive pruning and achieves more balanced performance across the two language conditions, obtaining 37.37% and 38.43% Hit@10 on English and Banglish, respectively. The code and dataset are publicly available at \href{https://github.com/M-Jawad01/Conversational-Job-Recommendation-System-LLM}{GitHub} and \href{https://huggingface.co/datasets/Armans33115/JobCCC-Conversational-Job-Recommendation-Bangladesh}{Hugging Face}, respectively.

Problem

• conversational job recommendation
• code-mixed low-resource settings
• constraint matching
• benchmark dataset

Innovation

• Conversational Job Recommendation
• Code-Mixed Low-Resource
• Soft-Constraint-Aware Ranking
• TOPSIS
• Multi-criteria Ranking