RESEARCH

Papers.

My thesis and the papers from my projects.

2026 · MASTER'S THESIS · ERICSSON
When Agentic Workflows Help: Hybrid Retrieval for Test Case Recommendation over Heterogeneous Software Artifacts
Berkay Orhan
This master's thesis designs, implements, and evaluates an agentic hybrid retrieval assistant for test scope analysis in a telecommunications setting. The system combines dense semantic retrieval, sparse lexical matching, structured relational queries, and bounded graph navigation over heterogeneous industrial test artifacts. The evaluation uses 52 hand-built scenarios over a corpus of 11,366 semi-structured industrial test cases containing free-text descriptions, test instructions, structured metadata, and trace links. Each scenario includes expected artifact identifiers, expected entity references, a reference answer, and an answerability label, and is run across ReAct and Orchestrator-worker workflows with full-prompt and skill-based setups.
Read the paper →Benchmark →
2025 · PROJECT PAPER
SynGraph: Deep Research Agent
Berkay Orhan
SynGraph documents a LangGraph-based research agent that routes user prompts, clarifies ambiguity, orchestrates parallel researcher workers, and synthesizes cited briefs. The stack pairs Gemini models for reasoning and routing with dual web/semantic search (Tavily + Exa) to keep outputs grounded.
Read the paper →Project →
2024 · PROJECT PAPER
AniMatch: A Content-Based Anime Recommendation System
Berkay Orhan, Jonatan Ebenholm
AniMatch is a content-based recommendation system designed to suggest anime titles based on their inherent features. The system employs BERT embeddings to extract meaningful representations of anime metadata and cosine similarity to compute relationships between titles. Unlike traditional recommendation systems, AniMatch focuses on metadata analysis, ensuring accurate and contextually relevant recommendations.
Read the paper →Project →

When Agentic Workflows Help: Hybrid Retrieval for Test Case Recommendation over Heterogeneous Software Artifacts

Berkay Orhan

This master's thesis designs, implements, and evaluates an agentic hybrid retrieval assistant for test scope analysis in a telecommunications setting. The system combines dense semantic retrieval, sparse lexical matching, structured relational queries, and bounded graph navigation over heterogeneous industrial test artifacts. The evaluation uses 52 hand-built scenarios over a corpus of 11,366 semi-structured industrial test cases containing free-text descriptions, test instructions, structured metadata, and trace links. Each scenario includes expected artifact identifiers, expected entity references, a reference answer, and an answerability label, and is run across ReAct and Orchestrator-worker workflows with full-prompt and skill-based setups.

SynGraph: Deep Research Agent

Berkay Orhan

SynGraph documents a LangGraph-based research agent that routes user prompts, clarifies ambiguity, orchestrates parallel researcher workers, and synthesizes cited briefs. The stack pairs Gemini models for reasoning and routing with dual web/semantic search (Tavily + Exa) to keep outputs grounded.

AniMatch: A Content-Based Anime Recommendation System

Berkay Orhan, Jonatan Ebenholm

AniMatch is a content-based recommendation system designed to suggest anime titles based on their inherent features. The system employs BERT embeddings to extract meaningful representations of anime metadata and cosine similarity to compute relationships between titles. Unlike traditional recommendation systems, AniMatch focuses on metadata analysis, ensuring accurate and contextually relevant recommendations.