PulseAugur
EN
LIVE 12:47:54

Semantic Search outperforms Deep Agentic Search in code QA, study finds

By PulseAugur Editorial · [1 sources] ·

A new study published on arXiv investigates two primary methods for code question answering within large repositories: Semantic Search and Deep Agentic Search. Researchers found that Semantic Search outperformed Deep Agentic Search, correctly answering 65.2% of questions compared to 46.2%, and did so at less than half the cost. The study identified that Deep Agentic Search, despite being a preferred design for protecting agents from context pollution, introduced a significant failure mode where sub-agents provided incorrect answers confidently. AI

IMPACT This research suggests that current practices in code agent design may be suboptimal, potentially impacting the efficiency and accuracy of AI-assisted software development.

RANK_REASON Research paper published on arXiv detailing an empirical study comparing two code question answering methods. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.IR (Information Retrieval) →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

Semantic Search outperforms Deep Agentic Search in code QA, study finds

COVERAGE [1]

  1. arXiv cs.IR (Information Retrieval) TIER_1 English(EN) · Mehmet S. Aktas ·

    Deep Agentic Search for Repository-Level Code Question Answering: An Empirical Study

    Code agents spend much of their effort simply locating the right code inside a repository. Two approaches dominate current practice. In Semantic Search, the agent retrieves code blocks from a vector index built from the repository in advance. In Deep Agentic Search (also known as…

RELATED ENTITIES

RELATED TOPICS