KGCE: Knowledge-Augmented Dual-Graph Evaluator for Cross-Platform Educational Agent Benchmarking with Multimodal Language Models

TL;DR

KGCE tackles the challenge of cross-platform educational agents by integrating a knowledge-base augmentation strategy with a dual-graph evaluator. It builds a 104-task education-focused dataset across Windows and Android, supported by a domain-specific knowledge base that informs multi-device workflows. The evaluation framework comprises a Task Completeness Graph and an Execution Efficiency Graph, enabling fine-grained metrics beyond coarse success criteria. Experiments across multiple MLMs show that knowledge augmentation consistently boosts task completion and efficiency, with GPT-4o achieving the best performance; findings highlight the practical impact for deploying educational agents in private-domain software and point to scalability and error-detection enhancements as future directions.

Abstract

With the rapid adoption of multimodal large language models (MLMs) in autonomous agents, cross-platform task execution capabilities in educational settings have garnered significant attention. However, existing benchmark frameworks still exhibit notable deficiencies in supporting cross-platform tasks in educational contexts, especially when dealing with school-specific software (such as XiaoYa Intelligent Assistant, HuaShi XiaZi, etc.), where the efficiency of agents often significantly decreases due to a lack of understanding of the structural specifics of these private-domain software. Additionally, current evaluation methods heavily rely on coarse-grained metrics like goal orientation or trajectory matching, making it challenging to capture the detailed execution and efficiency of agents in complex tasks. To address these issues, we propose KGCE (Knowledge-Augmented Dual-Graph Evaluator for Cross-Platform Educational Agent Benchmarking with Multimodal Language Models), a novel benchmarking platform that integrates knowledge base enhancement and a dual-graph evaluation framework. We first constructed a dataset comprising 104 education-related tasks, covering Windows, Android, and cross-platform collaborative tasks. KGCE introduces a dual-graph evaluation framework that decomposes tasks into multiple sub-goals and verifies their completion status, providing fine-grained evaluation metrics. To overcome the execution bottlenecks of existing agents in private-domain tasks, we developed an enhanced agent system incorporating a knowledge base specific to school-specific software. The code can be found at https://github.com/Kinginlife/KGCE.

Figures (4)

• Figure 1: The overall framework of KGCE. The system first generates tasks from educational datasets, then executes them through a pipeline of action prediction, execution, and evaluation. A dual-graph evaluator assesses task completeness and execution efficiency. Based on screenshot and OCR feedback, the system may invoke external knowledge from LLMs (e.g., GPT-4o, Qwen-VL, Gemini) to support cross-environment agents (Windows and Android) in accomplishing complex tasks.
• Figure 2: The knowledge base module. Given a task description and package names, the system identifies relevant packages, retrieves their descriptions from the knowledge base, and constructs prompts for the LLM. The knowledge base is organized by packages, pages, and elements, with each element including its position, description, and sub-elements.
• Figure 3: RQ1 Correlation of Metrics.
• Figure 4: RQ3 Comparison of Multimodal Large Language Models