2018/02/12 by Karamollah Bagherifard, Mohsen Rahmani, Vahid Rafe +1
Computer Science · #Artificial intelligence #Complementarity (molecular biology) #Computer science #Data mining #Information retrieval #Ontology #Recommender Systems and Techniques #Recommender system #Semantic Web #Semantic similarity #Similarity (geometry) #Taxonomy (biology) #Text and Document Classification Technologies #Web Data Mining and Analysis
paper · doi:10.1142/s0219649218500107
openalex publication_date 2018/02/12 · crossref created 2018/02/12 · crossref issued 2018/03/01 · crossref published 2018/03/01 · crossref published-print 2018/03/01 · crossref published-online 2018/03/09 · crossref deposited 2019/08/06 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/22 · crossref indexed 2026/08/06
Products and web pages are the main components of the e-commerce data knowledge and the relationship among them is an important issue to be highly considered in recommender systems. This study aims to focus on the similarity and complementarity relationships among the products that have wide applications in the recommender systems. In the previously proposed methods, products and their relationships were revealed using taxonomy and “IS-A” relationship. In addition, the similarity and complementarity calculations were conducted based on edge computation by assigning a similar degree to any edge. More specifically, the children of a concept in the taxonomy was supported by a similar father’s “IS-A” degree. In contrast, this study provides a new approach based on ontology, data mining, and automatic discovering algorithms for the relationships with different degrees for the edges among the concepts. Accordingly, these relationships are initialised according to the “IS-A” degree. With regard to this weighted taxonomy, the semantic similarity and complementarity are measured based on concept distance. In addition, the proposed recommender system is item-based, which uses semantic similarity and complementarity. The required data for the present study were collected from construction materials supplier. The results illustrated that our proposed method is effective for construction materials recommendation.