We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computation and Language

Title: Using virtual edges to extract keywords from texts modeled as complex networks

Abstract: Detecting keywords in texts is important for many text mining applications. Graph-based methods have been commonly used to automatically find the key concepts in texts, however, relevant information provided by embeddings has not been widely used to enrich the graph structure. Here we modeled texts co-occurrence networks, where nodes are words and edges are established either by contextual or semantical similarity. We compared two embedding approaches -- Word2vec and BERT -- to check whether edges created via word embeddings can improve the quality of the keyword extraction method. We found that, in fact, the use of virtual edges can improve the discriminability of co-occurrence networks. The best performance was obtained when we considered low percentages of addition of virtual (embedding) edges. A comparative analysis of structural and dynamical network metrics revealed the degree, PageRank, and accessibility are the metrics displaying the best performance in the model enriched with virtual edges.
Subjects: Computation and Language (cs.CL); Social and Information Networks (cs.SI)
Cite as: arXiv:2205.02172 [cs.CL]
  (or arXiv:2205.02172v1 [cs.CL] for this version)

Submission history

From: Diego Amancio [view email]
[v1] Wed, 4 May 2022 16:43:03 GMT (14034kb,D)

Link back to: arXiv, form interface, contact.