The documents distributed by this server have been provided by the contributing authors as a means to ensure timely dissemination of scholarly and technical work on a non-commercial basis. Copyright and all rights therein are maintained by the authors or by other copyright holders, not withstanding that they have offered their works here electronically. It is understood that all persons copying this information will adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.

Unsupervised Query-based Document Recommendation Using Deep Learning

Author:Wael Alkhatib, Steffen Schnitzer, Tim Steuer, Christoph Rensing
Date:April 2019
Kind:Article - use for journal articles only
Book title:20th International Conference on Computational Linguistics and Intelligent Text Processing (accepted for publication)
Keywords:content-based recommender system; word embbedings; deep learning; DSSM; ontologies
Research Area(s):Knowledge Media
Abstract:Traditional recommender systems, which aim to recommend documents, rely on statistical information only (collaborative filtering approaches) or on meta-information describing the user profile, the documents as well as their relations. They poorly consider the text semantics and often suffer from homophily by recommending documents which are similar to the documents already acquired by the user. In this work, we propose an unsupervised query-based recommender system for the recommendation of textual documents. The system relies on Deep Semantic Similarity Model (DSSM) to implicitly measure the semantic similarity between the user’s interest, represented the user’s history or interests as a query of keywords, and the available documents. Our approach uses an automatically generated ontology. This ontology is used to formulate the queries and to interpret the semantic relatedness between user preferences in the user model and the concepts representing the documents. The experimental results show that our system significantly outperforms the baseline in terms of FScore.

If the paper is not available from this page, you might contact the author(s) directly via the "People" section on our KOM Homepage.

[Export this entry to BibTeX]