Deji Zhao

Deji Zhao

Auto-generated profile, based on publicly available information

Deji Zhao

School of Computer Science and Engineering, Northeastern University, Shenyang, Liaoning, China

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 October 2024
Verified
Conference Paper
Open access
DZDeji Zhao
Deji Zhao
DHDong Hong Han
Dong Hong Han
YYYe Yuan
Ye Yuan
Bo NingBo Ning
Bo Ning
MLMengxiang Li
Mengxiang Li
SSShuangyong Song
Shuangyong Song
Open-domain multi-modal dialogue system heavily relies on visual information to generate contextually relevant responses. The existing open-domain multi-modal dialog generation methods ignore the complementary relationship between multiple modalities, and are difficult to integrate with LLMs. To tackle these challenges, we introduce AutoGraph, an innovative method for constructing visual context graphs automatically. We aim to structure complex information and seamlessly integrate it with large language models (LLMs), aligning information from multiple modalities at both semantic and structural levels. Specifically, we fully connect the text graphs and scene graphs, and then trim unnecessary edges via LLMs to automatically construct a visual context graph. Next, we design several graph sampling grammar for the first time to convert graph structures into sequence which is suitable for LLMs. Finally, we propose a two-stage fine-tuning strategy to allow LLMs to understand graph sampling grammar and generate responses. We validate our proposed method on text-based LLMs, and visual-based LLMs, respectively. Experimental results show that our proposed method achieves state-of-the-art performance on multiple public datasets.

30 October 2025 in Neural Networks
Verified
Journal Article
Closed access
DZDeji Zhao
Deji Zhao
DHDong Hong Han
Dong Hong Han
YYYe Yuan
Ye Yuan
Bo NingBo Ning
Bo Ning
YLYongxiang Li
Yongxiang Li
SSShuangyong Song
Shuangyong Song

23 December 2020 in Complexity
Verified
Journal Article
Open access
PDF file
Bo NingBo Ning
Bo Ning
YSYunhao Sun
Yunhao Sun
DZDeji Zhao
Deji Zhao
WXWeikang Xing
Weikang Xing
GLGuanyu Li
Guanyu Li
Subgraph matching on a large graph has become a popular research topic in the field of graph analysis, which has a wide range of applications including question answering and community detection. However, traditional edge-cutting strategy destroys the structure of indivisible knowledge in a large RDF graph. On the premise of load-balancing on subgraph division, a dominance-partitioned strategy is proposed to divide a large RDF graph without compromising the knowledge structure. Firstly, a dominance-connected pattern graph is extracted from a pattern graph to construct a dominance-partitioned pattern hypergraph, which divides a pattern graph as multiple fish-shaped pattern subgraphs. Secondly, a dominance-driven spectrum clustering strategy is used to gather the pattern subgraphs into multiple clusters. Thirdly, the dominance-partitioned subgraph matching algorithm is designed to conduct all isomorphic subgraphs on a cluster-partitioned RDF graph. Finally, experimental evaluation verifies that our strategy has higher time-efficiency of complex queries, and it has a better scalability on multiple machines and different data scales.