Home // DBKDA 2020, The Twelfth International Conference on Advances in Databases, Knowledge, and Data Applications // View article
Authors:
Ahmad Shahzad
Frans Coenen
Keywords: Graph Construction; Collaborative Filtering; Query Optimization; Normalization; Cold Start.
Abstract:
Graph abstraction is an intuitive and effective approach for collaborative filtering as used in, for example, recommender engines. However, for many collaborative filtering applications, the transactional data is kept in a relational database and, through bespoke processes, is Exported, Transformed and Loaded (ETL) into a graph database where collaborative filtering algorithms can be applied. However, the ETL process requires knowledge of the source relational database, the target graph database and the application domain. The ETL process, therefore, tends to be expensive, non-0ptimised for graph queries and relies heavily on application domain knowledge and understanding of the property graph engine for the graph database. In this paper, a mechanism is presented whereby data in a relational format, which is normalised to 5th normal form, can be automatically converted to a graph database format, through an automated process. The presented evaluation demonstrates, using the recommendation engine application domain as an example, that the proposed mechanism is more efficient than comparable approaches to reduce the execution time required for collaborative filtering.
Pages: 20 to 26
Copyright: Copyright (c) IARIA, 2020
Publication date: September 27, 2020
Published in: conference
ISSN: 2308-4332
ISBN: 978-1-61208-790-0
Location: Lisbon, Portugal
Dates: from September 27, 2020 to October 1, 2020