FCA2VEC: Embedding Techniques for Formal Concept Analysis

Dürrschnabel, Dominik; Hanika, Tom; Stubbemann, Maximilian

doi:10.1007/978-3-030-93278-7_3

Cited by 1 publication

(8 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For two nodes from the same set that are not connected to the same node in the other set, it predicts whether there should be an unobserved node in the other set that is connected to both nodes. For example, in an author-paper network, for two authors that do not have a co-authorship, it predicts if they will have a new co-authorship in the future [ 6 , 7 ]. Both types of bipartite link prediction have attracted increasing attention for high practical values [ 3 , 8 , 9 ].…”

Section: Introductionmentioning

confidence: 99%

“…For example, in an author-paper network, a bi-clique represents a group of co-researchers and their publications. To capture the information of bi-cliques, research in [ 7 ] proposed object2vec and attribute2vec , which aims to embed the nodes of a bipartite network into a vector space based on their co-occurrence relationship in the maximal bi-cliques of the network. To achieve this, they used the method of formal concept analysis (FCA) [ 7 , 20 , 21 ].…”

Section: Introductionmentioning

confidence: 99%

“…To capture the information of bi-cliques, research in [ 7 ] proposed object2vec and attribute2vec , which aims to embed the nodes of a bipartite network into a vector space based on their co-occurrence relationship in the maximal bi-cliques of the network. To achieve this, they used the method of formal concept analysis (FCA) [ 7 , 20 , 21 ]. FCA is a method for learning rules from a binary relational knowledge base, which is strongly connected with bipartite networks and bi-cliques.…”

Section: Introductionmentioning

confidence: 99%

“…With such an equivalence, the co-occurrences of nodes in a maximal bi-clique can be converted to the co-occurrences of objects in a formal concept, which resembles the co-occurrences of words in a sentence. Thus, one can use the embedding models similar to the well-known word embedding model Word2Vec [ 22 , 23 ] to embed the nodes into vectors [ 7 ]. Besides such a Word2Vec-like embedding method, in [ 20 ], the authors proposed another method for embedding the nodes based on bi-clique information with the help of Bidirectional Long Short-Term Memory (Bi-LSTM) [ 24 ].…”

Section: Introductionmentioning

confidence: 99%

“…Besides such a Word2Vec-like embedding method, in [ 20 ], the authors proposed another method for embedding the nodes based on bi-clique information with the help of Bidirectional Long Short-Term Memory (Bi-LSTM) [ 24 ]. Both methods have shown good performances [ 7 , 20 ]. These two methods open up a novel strategy for bipartite link prediction, that is, to convert bipartite networks into formal contexts and use FCA to extract information on bi-cliques so that we can further process them with other methods.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

BERT4FCA: A method for bipartite link prediction using formal concept analysis and BERT

Peng,

Yang,

Yamamoto

2024

PLoS ONE

View full text Add to dashboard Cite

Link prediction in bipartite networks finds practical applications in various domains, including friend recommendation in social networks and chemical reaction prediction in metabolic networks. Recent studies have highlighted the potential for link prediction by maximal bi-cliques, which is a structural feature within bipartite networks that can be extracted using formal concept analysis (FCA). Although previous FCA-based methods for bipartite link prediction have achieved good performance, they still have the problem that they cannot fully capture the information of maximal bi-cliques. To solve this problem, we propose a novel method for link prediction in bipartite networks, utilizing a BERT-like transformer encoder network to enhance the contribution of FCA to link prediction. Our method facilitates bipartite link prediction by learning more information from the maximal bi-cliques and their order relations extracted by FCA. Experimental results on five real-world bipartite networks demonstrate that our method outperforms previous FCA-based methods, a state-of-the-art Graph Neural Network(GNN)-based method, and classic methods such as matrix-factorization and node2vec.

show abstract

Section: Introductionmentioning

confidence: 99%