Panagiotis Symeonidis

Tag recommendations based on tensor dimensionality reduction

Social tagging is the process by which many users add metadata in the form of keywords, to annotate and categorize information items (songs, pictures, web links, products etc.). Collaborative tagging systems recommend tags to users based on what tags other users have used for the same items, aiming to develop a common consensus about which tags best describe an item. However, they fail to provide appropriate tag recommendations, because: (i) users may have different interests for an information item and (ii) information items may have multiple facets. In contrast to the current tag recommendation algorithms, our approach develops a unified framework to model the three types of entities that exist in a social tagging system: users, items and tags. These data is represented by a 3-order tensor, on which latent semantic analysis and dimensionality reduction is performed using the Higher Order Singular Value Decomposition (HOSVD) technique. We perform experimental comparison of the proposed method against two state-of-the-art tag recommendations algorithms with two real data sets (Last.fm and BibSonomy). Our results show significant improvements in terms of effectiveness measured through recall/precision.

A Unified Framework for Providing Recommendations in Social Tagging Systems Based on Ternary Semantic Analysis

Social tagging is the process by which many users add metadata in the form of keywords, to annotate and categorize items (songs, pictures, Web links, products, etc.). Social tagging systems (STSs) can provide three different types of recommendations: They can recommend 1) tags to users, based on what tags other users have used for the same items, 2) items to users, based on tags they have in common with other similar users, and 3) users with common social interest, based on common tags on similar items. However, users may have different interests for an item, and items may have multiple facets. In contrast to the current recommendation algorithms, our approach develops a unified framework to model the three types of entities that exist in a social tagging system: users, items, and tags. These data are modeled by a 3-order tensor, on which multiway latent semantic analysis and dimensionality reduction is performed using both the higher order singular value decomposition (HOSVD) method and the kernel-SVD smoothing technique. We perform experimental comparison of the proposed method against state-of-the-art recommendation algorithms with two real data sets (Last.fm and BibSonomy). Our results show significant improvements in terms of effectiveness measured through recall/precision.

Transitive node similarity for link prediction in social networks with positive and negative links

Online social networks (OSNs) like Facebook, and Myspace recommend new friends to registered users based on local features of the graph (i.e. based on the number of common friends that two users share). However, OSNs do not exploit the whole structure of the network. Instead, they consider only pathways of maximum length 2 between a user and his candidate friends. On the other hand, there are global approaches, which detect the overall path structure in a network, being computationally prohibitive for huge-size social networks. In this paper, we define a basic node similarity measure that captures effectively local graph features. We also exploit global graph features introducing transitive node similarity. Moreover, we derive variants of our method that apply in signed networks. We perform extensive experimental comparison of the proposed method against existing recommendation algorithms using synthetic and real data sets (Facebook, Hi5 and Epinions). Our experimental results show that our FriendTNS algorithm outperforms other approaches in terms of accuracy and it is also time efficient. We show that a significant accuracy improvement can be gained by using information about both positive and negative edges.

Product recommendation and rating prediction based on multi-modal social networks

Online Social Rating Networks (SRNs) such as Epinions and Flixter, allow users to form several implicit social networks, through their daily interactions like co-commenting on the same products, or similarly co-rating products. The majority of earlier work in Rating Prediction and Recommendation of products (e.g. Collaborative Filtering) mainly takes into account ratings of users on products. However, in SRNs users can also built their explicit social network by adding each other as friends. In this paper, we propose Social-Union, a method which combines similarity matrices derived from heterogeneous (unipartite and bipartite) explicit or implicit SRNs. Moreover, we propose an effective weighting strategy of SRNs influence based on their structured density. We also generalize our model for combining multiple social networks. We perform an extensive experimental comparison of the proposed method against existing rating prediction and product recommendation algorithms, using synthetic and two real data sets (Epinions and Flixter). Our experimental results show that our Social-Union algorithm is more effective in predicting rating and recommending products in SRNs.

Social Tagging Recommender Systems

The new generation of Web applications known as (STS) is successfully established and poised for continued growth. STS are open and inherently social; features that have been proven to encourage participation. But while STS bring new opportunities, they revive old problems, such as information overload. Recommender Systems are well known applications for increasing the level of relevant content over the “noise” that continuously grows as more and more content becomes available online. In STS however, we face new challenges. Users are interested in finding not only content, but also tags and even other users. Moreover, while traditional recommender systems usually operate over 2-way data arrays, STS data is represented as a third-order tensor or a hypergraph with hyperedges denoting (user, resource, tag) triples. In this chapter, we survey the most recent and state-of-the-art work about a whole new generation of recommender systems built to serve STS.We describe (a) novel facets of recommenders for STS, such as user, resource, and tag recommenders, (b) new approaches and algorithms for dealing with the ternary nature of STS data, and (c) recommender systems deployed in real world STS. Moreover, a concise comparison between existing works is presented, through which we identify and point out new research directions.

Nearest-biclusters collaborative filtering based on constant and coherent values

Collaborative Filtering (CF) Systems have been studied extensively for more than a decade to confront the “information overload” problem. Nearest-neighbor CF is based either on similarities between users or between items, to form a neighborhood of users or items, respectively. Recent research has tried to combine the two aforementioned approaches to improve effectiveness. Traditional clustering approaches (k-means or hierarchical clustering) has been also used to speed up the recommendation process. In this paper, we use biclustering to disclose this duality between users and items, by grouping them in both dimensions simultaneously. We propose a novel nearest-biclusters algorithm, which uses a new similarity measure that achieves partial matching of users’ preferences. We apply nearest-biclusters in combination with two different types of biclustering algorithms—Bimax and xMotif—for constant and coherent biclustering, respectively. Extensive performance evaluation results in three real-life data sets are provided, which show that the proposed method improves substantially the performance of the CF process.

Treatment of tuberculous spondylitis with anterior stabilization and titanium cage

We retrospectively reviewed 12 patients with spinal tuberculosis of the thoracic and lumbar spine who had radical débridement, anterior decompression, interbody arthrodesis with an anterior interbody titanium cage, and autologous bone grafts, combined with a standardized perioperative antituberculous regimen. Their mean age was 55.1 years and they were observed for a mean of 65.3 months. Indications for surgery included epidural abscess, structural destruction with instability, progressive kyphosis, and/or neurologic deterioration. Kyphotic deformity was corrected from a mean of 24.6° (range, 15°-32°) to a mean of 10° (range, 4°-18°). Tuberculous infection was controlled and bony fusion was achieved in all patients. No recurrence of infection or construct failure was recorded. All patients were safely mobilized within the first postoperative week; back pain fully resolved in eight patients and improved in the remaining four. We conclude that radical débridement followed by anterior stabilization with a titanium cage and bone grafting is a reasonable alternative for tuberculous spondylitis requiring surgical treatment. It enables accurate and lasting deformity correction and provides adequate stability to allow early mobilization. The presence of a titanium cage in an area of mycobacterial infection did not preclude infection control or lead to recurrence.Level of Evidence: Therapeutic study. Level IV (case series). Please see Guidelines for Authors for a complete description of levels of evidence.

MoviExplain: a recommender system with explanations

Providing justification to a recommendation gives credibility to a recommender system. Some recommender systems (Amazon.com etc.) try to explain their recommendations, in an effort to regain customer acceptance and trust. But their explanations are poor, because they are based solely on rating data, ignoring the content data. Our prototype system MoviExplain is a movie recommender system that provides both accurate and justifiable recommendations.

From biological to social networks: Link prediction based on multi-way spectral clustering

Link prediction in protein-protein interaction networks (PPINs) is an important task in biology, since the vast majority of biological functions involve such protein interactions. Link prediction is also important for online social networks (OSNs), which provide predictions about who is a friend of whom. Many link prediction methods for PPINs/OSNs are local-based and do not exploit all network structure, which limits prediction accuracy. On the other hand, there are global approaches to detect the overall path structure in a network, being computationally prohibitive for huge-size PPINs/OSNs. In this paper, we enhance a previously proposed multi-way spectral clustering method by introducing new ways to capture node proximity in both PPINs/OSNs. Our new enhanced method uses information obtained from the top few eigenvectors of the normalized Laplacian matrix. As a result, it produces a less noisy matrix, which is smaller and more compact than the original one. In this way, we are able to provide faster and more accurate link predictions. Moreover, our new spectral clustering model is based on the well-known Bray-Curtis coefficient to measure proximity between two nodes. Compared to traditional clustering algorithms, such as k-means and DBSCAN, which assume globular (convex) regions in Euclidean space, our approach is more flexible in capturing the non-connected components of a social graph and a wider range of cluster geometries. We perform an extensive experimental comparison of the proposed method against existing link prediction algorithms and k-means algorithm, using two synthetic data sets, three real social networks and three real human protein data sets. Our experimental results show that our SpectralLink algorithm outperforms the local approaches, the k-means algorithm and another spectral clustering method in terms of effectiveness, whereas it is more efficient than the global approaches.

Development of an emission inventory system from transport in Greece

Abstract In Greece, to date, a detailed emission inventory database from transport is not available. The Greek Ministry for the Environmental Physical Planning and Public Works has financed an “Emission Inventory System from Transport” (EIST). The EIST aims to be the first detailed and well-structured national emission database from transport (road, rail, air and sea transport and also off-road activities), focusing on the best-input datasets available, the most advanced pollutant emission algorithms and the highest spatial and temporal resolutions. The EIST can also be used as a decision support system (DSS) for environmental planning and development, as it is possible to examine the environmental effects of various “emission scenarios” which result from the application of different environmental measures and policies. The EIST was developed in distributed software fashion technology for “Windows 2000”. Its architecture involves an integrated framework of GIS and RDBMS technology systems equipped with interactive communication capabilities between peripheral software tools. In the present paper, the structure of the system is presented with emphasis on the emission calculation methodology applied. In order to illustrate the functionalities of the system, emission results, primarily derived from the geographical information system, are presented and some initial conclusions regarding the emissions from the transport sector in Greece are drawn.