发表机构
East West University; University of California San Diego（东西大学; 加州大学圣地亚哥分校）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对低资源代码混合环境下的对话式职位推荐，提出JobCCC基准和W-SCAR排序框架，通过软约束感知避免严格过滤导致的职位误删，在英语和孟加拉语-英语混合语上分别取得37.37%和38.43%的Hit@10。
AI 中文摘要
对话式职位推荐需要联合建模语义相关性、用户偏好、资格要求以及现实世界职业讨论中使用的嘈杂语言。这些挑战在低资源、代码混合的环境中尤为突出，因为严格的约束匹配可能会错误地排除原本合适的职位。我们推出了JobCCC，一个针对孟加拉国的对话式职位推荐基准，包含22,410条结构化职位发布和988个源自区域Reddit社区的多轮职业咨询对话。每个对话都标注了不断变化的求职者偏好，并链接到一个真实职位，同时在语义等价的英语和罗马化孟加拉语-英语变体中进行评估。我们将稀疏BM25检索、多语言密集检索及其硬约束过滤版本与加权软约束感知排序（W-SCAR）进行比较，W-SCAR是我们的多标准排序框架，结合了词汇相关性、语义相关性以及基于理想解相似性偏好排序技术（TOPSIS）的经验、地点、教育和薪资的分级效用。实验表明，严格过滤会持续降低检索性能，因为不完整的提取和脆弱的属性匹配会不可逆地移除相关职位。W-SCAR避免了破坏性剪枝，并在两种语言条件下取得了更均衡的性能，在英语和孟加拉语-英语混合语中分别获得了37.37%和38.43%的Hit@10。代码和数据集分别公开在GitHub和Hugging Face上。
英文摘要
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.
CommentsSubmitted to WSDM 2027
