Keyword-based Research Field Discovery with External Knowledge Aware Hierarchical Co-clustering

-- Views

November 04, 23

スライド概要

Kai Sugahara, Kazushi Okamoto: Keyword-based research field discovery with external knowledge aware hierarchical co-clustering, The 8th International Workshop on Advanced Computational Intelligence and Intelligent Informatics (IWACIII2023), 2023.11, Beijing, PRC.

profile-image

Data Science Research Group, The University of Electro-Communications

シェア

またはPlayer版

埋め込む »CMSなどでJSが使えない場合

関連スライド

各ページのテキスト
3.

Not Focusing Organized in two types: Topology based Focusing Topic based

4.

Document-term matrix Documents SVD Low-dimensional matrix Embedder Documents (based on DL) Embeddings -means Clusters Clustering Clusters

5.

Same meaning Word 0 Word 1 Word 2 Same meaning Word 3 Detect & Marge Word 0 Word 1' Word 4 Word 5 Utilizing knowledge regaring words Word 3 Word 4' Corpus

6.

All Papers Field A Field B Isn't that a bit too abstract? Field C All Papers Field D Field A Field B This granularity is good for my understanding! Field C Field C-1 Field D Field C-2

7.

Column 2 Column 3 Column 4 Column 4 1 0 Row 1 1 1 0 0 0 Row 2 0 0 1 1 Row 3 1 1 0 Row 4 0 0 1 cluster Column 1 Column 3 1 cluster Column 0 Column 2 0 Row 0 0 0 1 1 0 Row 2 0 0 1 1 0 0 Row 4 0 0 1 1 1 0 0 Row 1 1 1 0 0 0 1 1 Row 3 1 1 0 0 0 co- clustering cluster Column 1 0 cluster Column 0 Row 0 cluster

8.

Augmented Matrix Transfer knowledge Relational Matrix Column objects Row objects Transfer knowledge regarding row objects features Column objects Row objects features Augmented Matrix regarding column objects

10.

1. Data Preparation Citation Data Paper Data : Auxiliary Embeddings of papers abstracts and keywords of abstracts and keywords 2. Generation of Base Clusters 3. Generation of Matrices Construction of Citation Graph Relational Matrix Spectral Clustering Augmented Matrices regarding abstract and kw. 4. Generation of Clusters (for all parameter settings) Paper Clusters Clustering by HICCAM Calculation of Consensus Degree Keyword Clusters Paper Clusters (Base Clusters) Choose the Best in Concensus Degree Visualization of Clusters Keyword Clusters Paper Clusters Parameter Setteing

11.

papers keywords

13.

Transfer keyword knowledge Influence regarding papers Influence Relational Matrix Keywords Papers Transfer paper (abstract ) knowledge Augmented Matrix 350 - dimensions Keywords Papers 350 - dimensions Augmented Matrix regarding keywords

14.

B Base Clusters score paper A Spectral Clustering C E D A B C D E