InfluMatch: Frontier-Quality KOL Search at 4B-Model Cost
Quick Answer
InfluMatch introduces a cost-effective KOL search system using a three-stage cascade of small open-weight models, achieving 94.1% P@5 on an 11-query set.
Quick Take
InfluMatch introduces a cost-effective KOL search system using a three-stage cascade of small open-weight models, achieving 94.1% P@5 on an 11-query set. This system outperforms traditional methods and matches the frontier model Kimi-K2.6 while using 35x fewer output tokens and processing 50 KOL queries in approximately 20 seconds on an A100 GPU.
Key Points
- InfluMatch uses a retrieval, rerank, and reason cascade with small models.
- Achieves 94.1% P@5, outperforming traditional keyword searches.
- Processes 50 KOL queries in about 20 seconds on an A100 GPU.
- Utilizes 35x fewer output tokens than the frontier model Kimi-K2.6.
- Only pairwise fine-tuning improves reranker accuracy, not the reasoner.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Matching influencers (KOLs) to free-form, multi-part Thai marketing criteria is today served either by keyword search over structured profiles, which misses semantic fit, or by prompting frontier LLMs over every candidate, which is accurate but slow and expensive. We present InfluMatch, a low-cost three-stage cascade -- retrieval $\rightarrow$ rerank $\rightarrow$ reason -- built entirely from small open-weight models: dense retrieval returns 50 candidates, a 4B pointwise reranker scores each by the log-probability of a single Yes token and keeps 10, and a 4B reasoner grades the shortlist per criterion on a rubric with a Thai rationale. The cascade is designed for cost: reasoning over a filtered top-10 halves token spend versus reasoning over all 50 while scoring 14 points higher. End-to-end against human relevance labels on an 11-query set with all 50 candidates labeled, the full cascade reaches 94.1% P@5, versus a retrieval-only baseline near random; it matches the frontier model Kimi-K2.6 (91.8%) while emitting ${\sim}35\times$ fewer output tokens and serving a 50-KOL query in ${\sim}20$ s on one A100. Notably, the only fine-tuning that pays off is pairwise: a SimPO-tuned reranker matches the frontier baseline's best-pick accuracy (78.0 EM), whereas fine-tuning the reasoner on pointwise per-criterion labels improves offline scores yet degrades end-to-end ranking -- an inversion we trace to the design of the absolute labeling task -- leaving the untuned base model as the strongest deployed reasoner. The result is a deployable, explainable KOL search system at a small fraction of frontier serving cost.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR) |
| Cite as: | arXiv:2607.05968 [cs.CL] |
| (or arXiv:2607.05968v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2607.05968 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Krittanon Kaewtawee [view email]
[v1]
Tue, 7 Jul 2026 08:04:05 UTC (1,435 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CL
See more →Quantifying Prior Dominance in Systems
The study introduces the Normalized Context Utilization (NCU) metric to evaluate Retrieval-Augmented Generation (RAG) systems, revealing that Small Language Models (SLMs) outperform larger models in factual extraction. The findings indicate that traditional scaling laws yield diminishing returns, with a commercial API frequently failing against adversarial evidence due to systemic confidence collapse.
