Marco de Gemmis is Assistant Professor at the Department of Computer Science, University of Bari Aldo Moro, Italy, where he received his PhD in Computer Science in 2005.
His primary research interests include content-based recommender systems, natural language processing, information retrieval, text mining, and in general personalized information filtering.
He authored over 100 scientific articles published in international journals and collections, proceedings of international conferences and workshops, and book chapters.
He was program committee member for international conferences, including: ACM Recommender Systems; User Modeling, Adaptation and Personalization (UMAP), and served as a reviewer for international journals, including: User Modeling and User Adapted Interaction; ACM Transactions on Internet Technologies.
He was invited speaker at several universities, including: University of Roma 3, University of Basque Country San Sebastian, University of Cagliari, University of Milano-Bicocca, University of Naples Federico II.
Google Scholar profile: Marco de Gemmis @ Google Scholar
Contact him at email@example.com
Deep content analytics methods to improve transparency and serendipity of recommender systems
Advanced methods for Natural Language Processing and the availability of open knowledge sources, such as Wikipedia and BabelNet, have promoted recent progress in the field of content- based recommender systems (CBRSs).
Those systems analyze both item descriptions (content) and user ratings to infer user profiles, which store information about preferences, exploited to suggest items similar to those users liked in the past.
Novel research works have introduced Deep Content Analytics methods, i.e. semantic techniques that allow a better understanding of item properties, in terms of concepts instead of keywords. CBRSs can benefit from these techniques to implement more advanced, meaningful representations of items and user profiles.
The talk will provide a basic survey of semantic techniques:
- top-down approaches, based on the use of different open knowledge sources (ontologies,Wikipedia, DBpedia, BabelNet)
- bottom-up approaches, based on the distributional hypothesis, which states that "words that occur in the same contexts tend to have similar meanings"
The talk will show a semantic approach designed to provide explanations of suggestions, and a method for the discovery of hidden correlations among items, exploited to find "serendipitous" recommendations, i.e. items which are unexpected and potentially interesting at the same time.
14.30 - 16:10 - Session 1:
14.30 - 14.45 - Welcome Message - Opening remarks;
14.45 - 15.30 - Marco De Gemmis (Keynote);
- 15.30 - 15.50 - Context-specific Recommendation System for Predicting Similar PubMed Articles;
- 15.50 - 16.10 - Exploiting a Determinant-based Metric to Evaluate a Word-embeddings Matrix of Items;
16.10 - 16:30 - Coffe break
16.30 - 18:10 - Session 2:
- 16.30 - 16.50 - SeBPR: Semantics enhanced Bayesian Personalized Ranking with Comparable Item Pairs;
- 16.50 - 17.10 - Semantic Enabled Recommender System for Micro-blog users;
- 17.10 - 17.30 - SemStim: Exploiting Knowledge Graphs for Cross-Domain Recommendation;
17.30 - 18.10 - Open Discussion - Closing remarks.
Call for Papers:
A recommender system is designed to suggest items that are expected to interest a user. In order to filter the items and produce the recommendation, Data Mining techniques are largely employed. Among the most popular recommendation approaches in the literature and in real-world applications (e.g., e-commerce websites) are the so-called content-based recommender systems. Content-based recommender systems suggest to users items that are similar to those they previously evaluated . The early systems used relatively simple retrieval models, such as the Vector Space Model, with the basic TF-IDF weighting.
Simple (word-based) interest descriptions may fall short both because of semantic ambiguity and because they lack of generality. Recently, content-based recommender systems evolved and started employing external knowledge sources (e.g., ontologies) to improve accuracy and scope of recommendations [1,2].
More recent approaches have been based on deep learning . Other approaches, such as , have employed word embeddings in the recommendation process. Among the best known and high-performance implementations following these lines of research we mention Google's word2vec.
Given the rapid advances of Semantic Technologies, there is still a large number of options for recommender systems to take advantage of semantics.
Our workshop will solicit contributions in all topics related to employing Semantic Technologies in Recommender Systems, focused (but not limited) to the following list:
- Novel approaches to user profiling in recommender systems that model behavior with semantic technologies;
- Content-based recommendation algorithms that employ novel uses of semantic technologies;
- Recommendation explanation using semantic technologies;
- Generation of novel, diverse, and serendipitous recommendations using semantic technologies;
- Hybrid recommender systems that combine semantic technologies with other recommendation techniques (e.g, collaborative);
- Group-based approaches that use semantic technologies to describe the group preferences or to generate recommendations.
Accepted papers will be included in the IEEE ICDM 2016 Workshops Proceedings volume published by IEEE Computer Society Press, and will also be included in the IEEE Xplore Digital Library. The workshop proceedings will be in a CD separated from the CD of the main conference. The CD is produced by IEEE Conference Publishing Services (CPS).
 Marco de Gemmis, Pasquale Lops, Cataldo Musto, Fedelucio Narducci, and Giovanni Semeraro, “Semantics-aware content-based recommender systems,” in Recommender Systems Handbook, pp. 119–159. Springer US, 2015.
 Michel Capelle, Frederik Hogenboom, Alexander Hogenboom, and Flavius Frasincar, “Semantic news recommendation using wordnet and bing similarities,” in Proceedings of the 28th Annual ACM Symposium on Applied Computing, New York, NY, USA, 2013, pp. 296–302, ACM.
 Tomas Mikolov, Quoc V. Le, and Ilya Sutskever, “Exploiting similarities among languages for machine translation,” CoRR, vol. abs/1309.4168, 2013.
 Cataldo Musto, Giovanni Semeraro, Marco de Gemmis, and Pasquale Lops, “Learning word embeddings from wikipedia for content-based recommender systems,” in Advances in Information Retrieval - 38th European Conference on IR Research, ECIR 2016, Padua, Italy, March 20-23, 2016. Proceedings. 2016, vol. 9626 of Lecture Notes in Computer Science, pp. 729–734, Springer.
Types of contribution:
We will consider three different submission types, all in the IEEE 2-column format: regular (8 pages), short (4 pages) and extended abstracts (2 pages).
Research and position papers (regular or short) should be clearly placed with respect to the state of the art and state the contribution of the proposal in the domain of application, even if presenting preliminary results. In particular, research papers should describe the methodology in detail, experiments should be repeatable, and a comparison with the existing approaches in the literature should be made where possible. Position papers should introduce novel point of views in the workshop topics or summarize the experience of a researcher or a group in the field.
Practice and experience reports (short) should present in detail the real-world scenarios in which Semantic Technologies are employed for recommendation purposes.
Demo proposals (extended abstract) should present the details of a prototype or complete application that employs Semantic Technologies in Recommender Systems. The systems will be demonstrated to the workshop attendees.
All papers must be formatted according to the IEEE Computer Society proceedings manuscript style, following IEEE ICDM 2016 submission guidelines available at http://icdm2016.eurecat.org/.
Papers should be submitted in PDF format, electronically, using the CyberChair submission system, available at: ICDM-SERecSys@Cyberchair
For general enquires regarding the workshop, send an email to: firstname.lastname@example.org