Leveraging Diverse Text Embeddings and Mediums for Incorrect Assignment Detection
Keywords: Incorrect Assignment Detection, text embedding, GBDT, GNN
TL;DR: M1stic, 5th solution for whoiswho-kddcup-2024
Abstract: Abstract
Academic knowledge graph mining seeks to enhance our under-
standing of scientific evolution and trends, unlocking substantial
potential for guiding policy, facilitating talent discovery, and ad-
vancing knowledge acquisition. However, the field’s progress is
hindered by the absence of standardized benchmarks. To address
these issues through tasks focusing on name disambiguation com-
plexities and developing models to detect misattributed papers,
leveraging detailed paper attributes We used diverse text embed-
ding methods to extract semantic features of paper attributes, and
established an isomorphic graph structure based on the connections
between papers to capture potential associations between different
papers. By integrating the tree-base model and the graphsage model
achieved 5th place in WhoIsWho-IND-KDD-2024 competition
Submission Number: 31
Loading