Alejandro Corbellini

Persisting big-data: The NoSQL landscape

Abstract The growing popularity of massively accessed Web applications that store and analyze large amounts of data, being Facebook, Twitter and Google Search some prominent examples of such applications, have posed new requirements that greatly challenge traditional RDBMS. In response to this reality, a new way of creating and manipulating data stores, known as NoSQL databases, has arisen. This paper reviews implementations of NoSQL databases in order to provide an understanding of current tools and their uses. First, NoSQL databases are compared with traditional RDBMS and important concepts are explained. Only databases allowing to persist data and distribute them along different computing nodes are within the scope of this review. Moreover, NoSQL databases are divided into different types: Key-Value, Wide-Column, Document-oriented and Graph-oriented. In each case, a comparison of available databases is carried out based on their most important features.

An architecture and platform for developing distributed recommendation algorithms on large-scale social networks

The creation of new and better recommendation algorithms for social networks is currently receiving much attention owing to the increasing need for new tools to assist users. The volume of available social data as well as experimental datasets force recommendation algorithms to scale to many computers. Given that social networks can be modelled as graphs, a distributed graph-oriented support able to exploit computer clusters arises as a necessity. In this work, we propose an architecture, called Lightweight-Massive Graph Processing Architecture, which simplifies the design of graph-based recommendation algorithms on clusters of computers, and a Java implementation for this architecture composed of two parts: Graphly, an API offering operations to access graphs; and jLiME, a framework that supports the distribution of algorithm code and graph data. The motivation behind the creation of this architecture is to allow users to define recommendation algorithms through the API and then customize their execution using job distribution strategies, without modifying the original algorithm. Thus, algorithms can be programmed and evaluated without the burden of thinking about distribution and parallel concerns, while still supporting environment-level tuning of the distributed execution. To validate the proposal, the current implementation of the architecture was tested using a followee recommendation algorithm for Twitter as case study. These experiments illustrate the graph API, quantitatively evaluate different job distribution strategies w.r.t. recommendation time and resource usage, and demonstrate the importance of providing non-invasive tuning for recommendation algorithms.

Folksonomy-Based Recommender Systems: A State-of-the-Art Review

Collaborative tagging systems, also known as folksonomies, have grown in popularity over the Web on account of their simplicity to organize several types of content (e.g., Web pages, pictures, and video) using open‐ended tags. The rapid adoption of these systems has led to an increasing amount of users providing information about themselves and, at the same time, a growing and rich corpus of social knowledge that can be exploited by recommendation technologies. In this context, tripartite relationships between users, resources, and tags contained in folksonomies set new challenges for knowledge discovery approaches to be applied for the purposes of assisting users through recommendation systems. This review aims at providing a comprehensive overview of the literature in the field of folksonomy‐based recommender systems. Current recommendation approaches stemming from fields such as user modeling, collaborative filtering, content, and link‐analysis are reviewed and discussed to provide a starting point for researchers in the field as well as explore future research lines.

Exploring the role of personality traits in followee recommendation

– Followee recommendation is a problem rapidly gaining importance in Twitter as well as in other micro-blogging communities. To find interesting users to follow, most recommendation systems leverage different factors such as graph topology or user-generated content, among others. Those systems mostly disregard, however, the effect of psychological characteristics, such as personality, over the followee selection process. As personality is considered one of the primary factors that influence human behaviour, the purpose of this paper is to shed some light on the impact of personality traits on followee selection. , – The authors performed a data analysis comparing the similarity among Twitter users and their followees regarding personality traits. The authors analysed three different similarity measures. First, the authors computed an overall similarity considering the five personality traits or dimensions of the Five-Factor model as a whole. Second, the authors computed the dimension-to-dimension similarity considering each individual personality trait independently of each other. Third, the authors computed a cross-dimension similarity considering each personality dimension in relation to the others. , – This study showed that personality should be considered as a distinctive factor in the process of followee selection. However, personality dimensions should not be analysed as a whole as the overall personality similarity might not accurately assess the actual matching between individuals. Instead, the performed data analysis showed the existence of relations among the individual dimensions. Thus, the importance of considering each personality trait with respect to others is stated. , – This study is among the firsts to study the impact of personality, one of the primary factors that influence human behaviour and social relationships, in the selection of followees in micro-blogging communities.

DPM: A novel distributed large-scale social graph processing framework for link prediction algorithms

Abstract Large-scale graphs have become ubiquitous in social media. Computer-based recommendations in these huge graphs pose challenges in terms of algorithm design and resource usage efficiency when processing recommendations in distributed computing environments. Moreover, recommendation algorithms for graphs, particularly link prediction algorithms, have different requirements depending of the way the underlying graph is traversed. Path-based algorithms usually perform traversals in different directions to build a large ranking of vertices to recommend, whereas random walk-based algorithms build an initial subgraph and perform several iterations on those vertices to compute the final ranking. In this work, we propose a distributed graph processing framework called Distributed Partitioned Merge (DPM), which supports both types of algorithms and we compare its performance and resource usage w.r.t. two relevant frameworks, namely Fork-Join and Pregel. In our experiments, we show that in most tests DPM outperforms both Pregel and Fork-Join in terms of recommendation time, with a minor penalization in network usage in some scenarios.

Personality-aware followee recommendation algorithms

As the popularity of micro-blogging sites, expressed as the number of active users and volume of online activities, increases, the difficulty of deciding who to follow also increases. Such decision might not depend on a unique factor as users usually have several reasons for choosing whom to follow. However, most recommendation systems almost exclusively rely on only two traditional factors: graph topology and user-generated content, disregarding the effect of psychological and behavioural characteristics, such as personality, over the followee selection process. Due to its effect over peoples reactions and interactions with other individuals, personality is considered as one of the primary factors that influence human behaviour. This study aims at assessing the impact of personality in the accurate prediction of followees, beyond simple topological and content-based factors. It analyses whether user personality could condition followee selection by combining personality traits with the most commonly used followee predictive factors. Results showed that an accurate appreciation of such predictive factors tied to a quantitative analysis of personality is crucial for guiding the search of potential followees, and thus, enhance recommendations. HighlightsThe impact of personality in the accurate prediction of followees is assessed.Personality was quantitatively assessed and combined with common recommendation factors.The combination of predicted factors was inserted into a recommendation algorithm.Adding personality can significantly enhance recommendation precision.Personality should be considered as a distinctive factor in followee selection.

Intelligent analysis of user interactions in a collaborative software engineering context

Software engineering is inherently a collaborative and social activity. Collaborative software engineering is a research area that aims at providing computer-based support to developers in the form of tools for coordination, communication and management. In this context, we present Paynal, an application that provides a set of collaboration facilities integrated into a development environment. Most important, Paynal provides managers and project leaders tools based on social network analysis to discover knowledge about team members and their relationships. We describe a case study in which Paynal has been successfully evaluated.

A Programming Interface and Platform Support for Developing Recommendation Algorithms on Large-Scale Social Networks

Friend recommendation algorithms in large-scale social networks such as Facebook or Twitter usually require the exploration of huge user graphs. In current solutions for parallelizing graph algorithms, the burden of dealing with distributed concerns falls on algorithm developers. In this paper, a simple yet powerful programming interface (API) to implement distributed graph traversal algorithms is presented. A case study on implementing a followee recommendation algorithm for Twitter using the API is described. This case study not only illustrates the simplicity offered by the API for developing algorithms, but also how different aspects of the distributed solutions can be treated and experimented without altering the algorithm code. Experiments evaluating the performance of different job scheduling strategies illustrate the flexibility or our approach.

Task Scheduling for Processing Big Graphs in Heterogeneous Commodity Clusters

Large-scale graph processing is a challenging problem since vertices can be arbitrarily connected, reducing locality and easily expanding the solution space. As a result, in recent years, a new breed of distributed frameworks that handle graphs efficiently has emerged. In large clusters with many resources (RAM, CPUs, network connectivity), these frameworks focus on exploiting the available resources as efficiently as possible. However, on situations where the cluster hardware is unbalanced or low in computing resources, the framework must correctly allocate tasks in order to complete execution. In this work, we compare three frameworks, the generic Fork-Join framework adapted to graph processing, and the Pregel and DPM frameworks that were originally designed for computing graphs. A link-prediction algorithm was used as case study to analyze several scheduling strategies that allocate tasks to servers in a cluster of heterogeneous characteristics. The dataset used for the experiments is a snapshot from the Twitter graph, and specifically, a subset of its users that pushed the memory requirements of the algorithm.

An Evaluation of Distributed Processing Models for Random Walk-Based Link Prediction Algorithms Over Social Big Data

The problem of inferring missing relationships between people in online social networks such as Facebook, Google+ and Twitter is currently being given much attention due to its enormous applicability. To this end, link prediction algorithms which operate on graph data have been considered. However, the relentless increase of the size of such networks calls for distributed processing models able to cope with the associated big amounts of data. In this paper, we study the suitability of three models (Fork-Join, Pregel and DPM) for scaling up a common class of such algorithms, i.e. random walk-based. Broadly, Fork-Join and Pregel promote two rather different ways of creating and handling parallel sub-computations, while DPM is a model combining the best of both. Experiments performed with the Twitter graph and two classical random walk-based algorithms named HITS and SALSA show that DPM outperforms Fork-Join and Pregel by [30–40]% and [10–20]% respectively in terms of recommendation time.