Riccardo Guidotti

Interaction Prediction in Dynamic Networks exploiting Community Discovery

Due to the growing availability of online social services, interactions between people became more and more easy to establish and track. Online social human activities generate digital footprints, that describe complex, rapidly evolving, dynamic networks. In such scenario one of the most challenging task to address involves the prediction of future interactions between couples of actors. In this study, we want to leverage networks dynamics and community structure to predict which are the future interactions more likely to appear. To this extent, we propose a supervised learning approach which exploit features computed by time-aware forecasts of topological measures calculated between pair of nodes belonging to the same community. Our experiments on real dynamic networks show that the designed analytical process is able to achieve interesting results.

MyWay: Location prediction via mobility profiling

Abstract Forecasting the future positions of mobile users is a valuable task allowing us to operate efficiently a myriad of different applications which need this type of information. We propose MyWay, a prediction system which exploits the individual systematic behaviors modeled by mobility profiles to predict human movements. MyWay provides three strategies: the individual strategy uses only the user individual mobility profile, the collective strategy takes advantage of all users individual systematic behaviors, and the hybrid strategy that is a combination of the previous two. A key point is that MyWay only requires the sharing of individual mobility profiles, a concise representation of the user׳s movements, instead of raw trajectory data revealing the detailed movement of the users. We evaluate the prediction performances of our proposal by a deep experimentation on large real-world data. The results highlight that the synergy between the individual and collective knowledge is the key for a better prediction and allow the system to outperform the state-of-art methods.

TOSCA: two-steps clustering algorithm for personal locations detection

One of the key tasks in mobility data analysis is the study of the individual mobility of users with reference to their personal locations, i.e. the places or areas where they stop to perform any kind of activities. Correctly discovering such personal locations is therefore a very important problem, which is yet not very well addressed in literature. In this work we propose a robust, efficient, statistically well-founded and parameter-free personal location detection process. The algorithm, called TOSCA (TwO-Steps parameter free Clustering Algorithm), combines two clustering strategies and applies statistical tests to drive the selection of the needed parameters. The proposed solution is tested against a large set of competitors and several datasets, including synthetic and real ones. The empirical results show its ability to automatically adapt to different contexts yielding good accuracy and a good efficiency.

Never drive alone

Carpooling, i.e., the act where two or more travelers share the same car for a common trip, is one of the possibilities brought forward to reduce traffic and its externalities, but experience shows that it is difficult to boost the adoption of carpooling to significant levels. In our study, we analyze the potential impact of carpooling as a collective phenomenon emerging from peoples mobility, by network analytics. Based on big mobility data from travelers in a given territory, we construct the network of potential carpooling, where nodes correspond to the users and links to possible shared trips, and analyze the structural and topological properties of this network, such as network communities and node ranking, to the purpose of highlighting the subpopulations with higher chances to create a carpooling community, and the propensity of users to be either drivers or passengers in a shared car. Our study is anchored to reality thanks to a large mobility dataset, consisting of the complete one-month-long GPS trajectories of approx. 10% circulating cars in Tuscany. We also analyze the aggregated outcome of carpooling by means of empirical simulations, showing how an assignment policy exploiting the network analytic concepts of communities and node rankings minimizes the number of single occupancy vehicles observed after carpooling.

A Survey of Methods for Explaining Black Box Models

In recent years, many accurate decision support systems have been constructed as black boxes, that is as systems that hide their internal logic to the user. This lack of explanation constitutes both a practical and an ethical issue. The literature reports many approaches aimed at overcoming this crucial weakness, sometimes at the cost of sacrificing accuracy for interpretability. The applications in which black box decision systems can be used are various, and each approach is typically developed to provide a solution for a specific problem and, as a consequence, it explicitly or implicitly delineates its own definition of interpretability and explanation. The aim of this article is to provide a classification of the main problems addressed in the literature with respect to the notion of explanation and the type of black box system. Given a problem definition, a black box type, and a desired explanation, this survey should help the researcher to find the proposals more useful for his own work. The proposed classification of approaches to open black box models should also be useful for putting the many research open questions in perspective.

A supervised approach for intra-/inter-community interaction prediction in dynamic social networks

Due to the growing availability of Internet services in the last decade, the interactions between people became more and more easy to establish. For example, we can have an intercontinental job interview, or we can send real-time multimedia content to any friend of us just owning a smartphone. All this kind of human activities generates digital footprints, that describe a complex, rapidly evolving, network structures. In such dynamic scenario, one of the most challenging tasks involves the prediction of future interactions between couples of actors (i.e., users in online social networks, researchers in collaboration networks). In this paper, we approach such problem by leveraging networks dynamics: to this extent, we propose a supervised learning approach which exploits features computed by time-aware forecasts of topological measures calculated between node pairs. Moreover, since real social networks are generally composed by weakly connected modules, we instantiate the interaction prediction problem in two disjoint applicative scenarios: intra-community and inter-community link prediction. Experimental results on real time-stamped networks show how our approach is able to reach high accuracy. Furthermore, we analyze the performances of our methodology when varying the typologies of features, community discovery algorithms and forecast methods.

Behavioral entropy and profitability in retail

Human behavior is predictable in principle: people are systematic in their everyday choices. This predictability can be used to plan events and infrastructure, both for the public good and for private gains. In this paper we investigate the largely unexplored relationship between the systematic behavior of a customer and its profitability for a retail company. We estimate a customers behavioral entropy over two dimensions: the basket entropy is the variety of what customers buy, and the spatio-temporal entropy is the spatial and temporal variety of their shopping sessions. To estimate the basket and the spatio-temporal entropy we use data mining and information theoretic techniques. We find that predictable systematic customers are more profitable for a supermarket: their average per capita expenditures are higher than non systematic customers and they visit the shops more often. However, this higher individual profitability is masked by its overall level. The highly systematic customers are a minority of the customer set. As a consequence, the total amount of revenues they generate is small. We suggest that favoring a systematic behavior in their customers might be a good strategy for supermarkets to increase revenue. These results are based on data coming from a large Italian supermarket chain, including more than 50 thousand customers visiting 23 shops to purchase more than 80 thousand distinct products.

Social or Green? A Data-Driven Approach for More Enjoyable Carpooling

Carpooling, i.e. the sharing of vehicles to reach common destinations, is often performed to reduce costs and pollution. Recent works on carpooling and journey planning take into account, besides mobility match, also social aspects and, more generally, non-monetary rewards. In line with this, we present a data-driven methodology for a more enjoyable carpooling. We introduce a measure of enjoyability based on peoples interests, social links, and tendency to connect to people with similar or dissimilar interests. We devise a methodology to compute enjoyability from crowd-sourced data, and we show how this can be used on real world datasets to optimize for both mobility and enjoyability. Our methodology was tested on real data from Rome and San Francisco. We compare the results of an optimization model minimizing the number of cars, and a greedy approach maximing the enjoyability. We evaluate them in terms of cars saved, and average enjoyability of the system. We present also the results of a user study, with more than 200 users reporting an interest of 39% in the enjoyable solution. Moreoever, 24% of people declared that sharing the car with interesting people would be the primary motivation for carpooling.

Towards user-centric data management: individual mobility analytics for collective services

We are under the big data microscope, and our digital traces are an inestimable source of awareness to deeply understand mobility phenomena as well as economic trends, social relationships and so on. Setting the focus of the big data microscope to capture human systematic behavior is surely a promising direction. The proposed vision is a methodological framework aimed to deal with intelligent personal data store that are able to automatically perform individual data mining, and that can provide proactive suggestions and support decisions, allow to share individual profiles in order to reach a level of knowledge comparable to those belonged to a collective system, and suggest interactions between individual and collective data mining in order to overtake the level of complex society knowledge extracted by the state-of-art methods. The study of individuals profiles, and the comparison and interactions with collective patterns, is dramatically helpful both for the novel detailed information retrieved through the methodological framework and for the possibility to deal at the same time with privacy issues.

Managing travels with PETRA: The Rome use case

The aim of the PETRA project is to provide the basis for a city-wide transportation system that supports policies catering for both individual preferences of users and city-wide travel patterns. The PETRA platform will be initially deployed in the partner city of Rome, and later in Venice, and Tel-Aviv.