Topical key concept extraction from folksonomy through graph-based ranking

• Published in: Multimedia tools and applications Vol. 75; no. 15; pp. 8875 - 8893
• Main Authors: Xue, Han, Qin, Bing, Liu, Ting
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.08.2016; Springer Nature B.V
• Subjects: Analysis; Computer Communication Networks; Computer Science; Data Structures and Information Theory; Extraction; Feature extraction; Graph theory; Information retrieval; Knowledge acquisition; Methods; Motion pictures; Multimedia; Multimedia computer applications; Multimedia Information Systems; Natural language processing; Ontology; Random walk theory; Ranking; Semantics; Social networks; Special Purpose and Application-Based Systems; Studies; Subject specialists; Tagging; Tags; Texts; User generated content; Vocabularies & taxonomies; Graph-based ranking; Topic-sensitive random walk; Folksonomy; Topical key concept extraction
• ISSN: 1380-7501; 1573-7721
• DOI: 10.1007/s11042-014-2303-9

Existing studies for concept extraction mainly focus on text corpora and indiscriminately mix numerous topics, which may lead to a knowledge acquisition bottleneck and misconception. We thus propose a novel method for extracting topical key concepts from folksonomy. This method can overcome the aforementioned problems through rich user-generated content and topic-sensitive concept extraction. We first identify topics from folksonomy by using topic models. Tags are then ranked according to importance relative to a certain topic through graph-based ranking. The top-ranking tags are extracted as topical key concepts. The combination of a novel edge weight and preference is proposed in tag importance propagation. The proposed method is applied to different datasets and is found to outperform the state-of-the-art baselines significantly. From the perspectives of parameter influence and case study, the proposed method is feasible and effective.