I published in peer-reviewed international journals (IEEE Transactions on Knowledge and Data Engineering (TKDE), World Wide Web Journal (WWW), Expert Systems with Applications (ESWA)) and conferences (MEDI, SMAP, MEDES, AC).
This page lists a selection of my most recent publication. For a complete list you can refer to my DBLP profile, to my Google Scholar profile, or to my Researchgate profile.
Recommender systems in location-based social networks (LBSNs), such as Facebook Places and Foursquare, have focused on recommending friends or locations to registered users by combining information derived from explicit (i.e. friendship network) and implicit (i.e. user-item rating network, user-location network, etc.) sub-networks. However, previous models were static and failed to adequately capture user time-varying preferences. In this paper, we provide a novel recommendation method based on the time dimension as well. We construct a hybrid tripartite (i.e., user, location, session) graph, which incorporates 7 different unipartite and bipartite graphs. Then, we test it with an extended version of the Random Walk with Restart (RWR) algorithm, which randomly walks through the network by using paths of 7 differently weighted edge types (i.e., user-location, user-session, user-user, etc.). We evaluate experimentally our method and compare it against three state-of-the-art algorithms on two real-life datasets; we show a significant prevalence of our method over its competitors.
Location-Based Social Networks (LBSNs) allow users to post ratings and reviews and to notify friends of these posts. Several models have been proposed for Point-of-Interest (POI) recommendation that use explicit (i.e. ratings, comments) or implicit (i.e. statistical scores, views, and user influence) information. However the models so far fail to capture sufficiently user preferences as they change spatially and temporally. We argue that time is a crucial factor because user check-in behavior might be periodic and time dependent, e.g. check-in near work in the mornings and check-in close to home in the evenings. In this paper, we present two novel unified models that provide review and POI recommendations and consider simultaneously the spatial, textual and temporal factors. In particular, the first model provides review recommendations by incorporating into the same unified framework the spatial influence of the users’ reviews and the textual influence of the reviews. The second model provides POI recommendations by combining the spatial influence of the users’ check-in history and the social influence of the users’ reviews into another unified framework. Furthermore, for both models we consider the temporal dimension and measure the impact of time on various time intervals. We evaluate the performance of our models against 10 other methods in terms of precision and recall. The results indicate that our models outperform the other methods.
Recommender systems elicit the interests and preferences of individuals and make recommendations accordingly, a main challenge for expert and intelligent systems. An essential problem in recommender systems is to learn users’ preference dynamics, that is, the constant evolution of the explicit or the implicit information, which is diversified throughout time according to the user actions. Also, in real settings data sparsity degrades the recommendation accuracy. Hence, state-of-the-art methods exploit multimodal information of users-item interactions to reduce sparsity, but they ignore preference dynamics and do not capture users’ most recent preferences. In this article, we present a Temporal Collective Matrix Factorization (TCMF) model, making the following contributions: (i) we capture preference dynamics through a joint decomposition model that extracts the user temporal patterns, and (ii) co-factorize the temporal patterns with multimodal user-item interactions by minimizing a joint objective function to generate the recommendations. We evaluate the performance of TCMF in terms of accuracy and root mean square error, and show that the proposed model significantly outperforms state-of-the-art strategies.
Recently, location-based social networks (LBSNs) gave the opportunity to users to share geo-tagged information along with photos, videos, and SMSs. Recommender systems can exploit this geographic information to provide much more accurate and reliable recommendations to users. In this paper, we present and compare 16 real life LBSNs, bringing into surface their advantages/ disadvantages, their special functionalities, and their impact in the mobile social Web. Moreover, we describe and compare extensively 43 state-of-the-art recommendation algorithms for LBSNs. We categorize these algorithms according to: personalization type, recommendation type, data factors/features, problem modeling methodology, and data representation. In addition to the above categorizations which cannot cover all algorithms in an integrated way, we also propose a hybrid k-partite graph taxonomy to categorize them based on the number of the involved k-partite graphs. Finally, we compare the recommendation algorithms with respect to their evaluation methodology (i.e., datasets and metrics) and we highlight new perspectives for future work in LBSNs.
Recommender systems in location-based social networks (LBSNs), such as Facebook Places and Foursquare, have focused on recommending friends or locations to registered users by combining information derived from explicit (i.e. friendship network) and implicit (i.e. user-item rating network, user-location network, etc.) sub-networks. However, previous’s work models were static, failing to capture adequately user preferences as they change over time. In this paper, we provide a novel recommendation method by incorporating the time dimension into our model through an auxiliary artificial node (i.e. session). In particular, we construct a hybrid tripartite (i.e., user, location, session) graph, which incorporates 7 different unipartite and bipartite graphs. Then, we run on it the well known Random Walk with Restart (RWR) algorithm, which randomly propagate through the network structure which has 7 differently weighted edge types (i.e., user-location, user-session, user-user, etc.) among its entities. We evaluate experimentally how RWR improve the procession of the recommendations during different time-windows against one state-of-the-art algorithm over the GeoSocialRec and the Foursquare datasets.
Nowadays, Online Social Networks have given the opportunity to users to share their interests. Moreover Location-Based Social Network added the location factor giving a new perspective to users' check-ins in POIs through smartphones. There are three main parameters characterizing these networks: mobility, proximity and periodicity. Here, we argue that periodicity is a significant upcoming trend in recommender systems. In particular, we present an extended comparison among 9 recommendation frameworks and their structural components. Moreover, we examine whether they provide personalized recommendations or not, the recommendation type they support, the data factors/features they use, the preferred methodology with which they model the problem and the data representation model they have chosen. By gathering this information we give an overview of the techniques and the features used and define new trends in this domain. The main factor is time that refines the final recommendation revealing relations among entities, which can increase accuracy of the proposals.
Online social networks have attracted users' attention in the last decade. Recommendation services constitute a critical functionality of such social platforms: users receive recommendations about resources (documents, pieces of music) and potential friends (people with the same interests). Recently, technological progressions in smart phones enabled the exploitation of geographical data information in social networks. Users can now receive recommendations about new Points of Interest (POIs), and new activities in POIs. Eventually, Location-based Social Networks (LBSNs) may become the 'Next Big Thing' of the Internet industry. This paper surveys the related work and current state-of-the-art algorithms in LBSNs. We also provide three new perspectives that concern recommendations in LBSNs: time-awareness, user's privacy issues, and explainability of recommendations. We present the latest work in LBSNs by comparing real systems and by categorizing them in multiple ways (platforms, personalization, etc.).
In this paper we propose a tool which has the ability to recognizes a subset o English language and translate them in to the corresponding SQL command. The system receives as input a query in a well specified subset of the English language and a database scheme. It processes this data accordingly to certain rules and acknowledgements in order to produce as output a correct query in sql language paper must have an abstract.
Logic in computer science describes topics where reasoning models are applied to the representation of knowledge and artificial intelligence. The use of distributed systems in solving logic problems is an intriguing but also challenging topic and algorithms of pruning network queries result in optimization of the resolution process. The current study suggests an implemented tool of a Distributed Peer-To-Peer Reasoning model, where local reasoning rules with incomplete facts are evaluated. This is achieved by sending queries of the neededfacts to peers, using a protocol that relieves the network from unnecessary queries and queries that linger and circle around.
Logic in computer science describes topics where logic is applied to the representation of knowledge and artificial intelligence, including digital circuit design, database systems, inference systems and more. The use of distributed systems, in solving logic problems, is an interesting and challenging topic in computer science, where algorithms of pruning network queries result in speeding up the resolution process. The current study suggests an implemented tool of a distributed peer-to-peer reasoning model, where local reasoning rules, with incomplete facts, are evaluated. This is achieved by sending queries of the needed facts to peers, using a protocol that relieves the network from unnecessary queries and queries that linger and circle around.
Many DB management systems use SQL syntax in order to provide a mechanism for data accessing. Although SQL is widely used, its syntax is not very comprehensive to a typical user. In this paper we present a tool capable of transforming simple queries formatted in natural language into their equivalent SQL queries. The purpose of this study is to provide a query syntax method closer to natural language. We currently support three languages: English, German and Greek and we show how our tool can be extended easily from a user so as to support any other language.
Recent years, the evolving nature of social networks has led to the expeditious growth of the internet and the rapid increment of the data on a global scale. The need of accessing and retrieving relevant information close to users' preferences is an open problem which continuously raises new challenges for recommender systems. To over come this problem many researchers focused on creating models that provide personalized recommendation in order to assist users making choices. In this context, a new research area in information systems has emerged over the last decades to respond to these challenges. This research area is called recommendation systems and focuses on modelling and analyzing data in order to retrieve relevant information based on users' preferences and to suggest some new alternatives. The exploitation of information in large amounts with the existing models is not sufficient since power law distribution of the data causes sparsity problem and makes personalization a difficult task. More specifically, models predictions accuracy is higher for users with large past history than for users with a few interactions. This problem is also known as "Cold Start" according to which it is attempted to retrieve relevant information based on small past history and to associate these users with other users, locations or products. In this problem, we should also examine the dynamics of users' preferences among time periods, which continuously alternate the precision of prediction models. On the other hand, there are models which consider users' preference dynamics but miss to personalize their recommendations since the past history information of most users is small. These approaches focus only on preference dynamics but lack the users' implicit and/or the explicit information. Thus, they can not adequately provide recommendations for users with short past history. Finally, there are approaches which increase their prediction robustness using geographical information but they miss the temporal dynamics and the side information related to the users. Most of the models in literature miss dynamics resulting to decrease their ability to personalize their recommendations and to solve the cold start problem. With this dissertation, I investigate the impact of all aforementioned issues on predictability of the models while retrieving relevant information to users' past history independent of the amount of the past history record. To this context, we present 5 novel recommendation models which improve the state-of-the-art approaches, and a novel model taxonomy based on the participant information networks.