The total knowledge contained within a collective supersedes the knowledge of even its most intelligent member. Yet the collective knowledge will remain inaccessible to us unless we are able to find efficient knowledge aggregation methods that produce reliable decisions based on the behavior or opinions of the collective’s members. It is often stated that simple averaging of a pool of opinions is a good and in many cases the optimal way to extract knowledge from a crowd. The method of averaging has been applied to analysis of decision-making in very different fields, such as forecasting, collective animal behavior, individual psychology, and machine learning. Two mathematical theorems, Condorcet’s theorem and Jensen’s inequality, provide a general theoretical justification for the averaging procedure. Yet the necessary conditions which guarantee the applicability of these theorems are often not met in practice. Under such circumstances, averaging can lead to suboptimal and sometimes very poor performance. Practitioners in many different fields have independently developed procedures to counteract the failures of averaging. We review such knowledge aggregation procedures and interpret the methods in the light of a statistical decision theory framework to explain when their application is justified. Our analysis indicates that in the ideal case, there should be a matching between the aggregation procedure and the nature of the knowledge distribution, correlations, and associated error costs. This leads us to explore how machine learning techniques can be used to extract near-optimal decision rules in a data-driven manner. We end with a discussion of open frontiers in the domain of knowledge aggregation and collective intelligence in general.

 

Rescuing Collective Wisdom when the Average Group Opinion Is Wrong

Andres Laan, Gabriel Madirolas, and Gonzalo G. de Polavieja

Front. Robot. AI, 06 November 2017 | https://doi.org/10.3389/frobt.2017.00056

Source: www.frontiersin.org

Falling oil revenues and rapid urbanization are putting a strain on the budgets of oil producing nations which often subsidize domestic fuel consumption. A direct way to decrease the impact of subsidies is to reduce fuel consumption by reducing congestion and car trips. While fuel consumption models have started to incorporate data sources from ubiquitous sensing devices, the opportunity is to develop comprehensive models at urban scale leveraging sources such as Global Positioning System (GPS) data and Call Detail Records. We combine these big data sets in a novel method to model fuel consumption within a city and estimate how it may change due to different scenarios. To do so we calibrate a fuel consumption model for use on any car fleet fuel economy distribution and apply it in Riyadh, Saudi Arabia. The model proposed, based on speed profiles, is then used to test the effects on fuel consumption of reducing flow, both randomly and by targeting the most fuel inefficient trips in the city. The estimates considerably improve baseline methods based on average speeds, showing the benefits of the information added by the GPS data fusion. The presented method can be adapted to also measure emissions. The results constitute a clear application of data analysis tools to help decision makers compare policies aimed at achieving economic and environmental goals.

 

Big Data Fusion to Estimate Fuel Consumption: A Case Study of Riyadh
Adham Kalila, Zeyad Awwad, Riccardo Di Clemente, Marta C. González

Source: arxiv.org

Historically, network science focused on static networks, in which nodes are connected by permanent links. However, in networked systems ranging from protein-protein interactions to social networks, links change. Although it might seem that permanent links would make it easier to control a system, Li et al. demonstrate that temporality has advantages in real and simulated networks. Temporal networks can be controlled more efficiently and require less energy than their static counterparts.

 

The fundamental advantages of temporal networks
A. Li, S. P. Cornelius, Y.-Y. Liu, L. Wang, A.-L. Barabási
Science  24 Nov 2017:
Vol. 358, Issue 6366, pp. 1042-1046
DOI: 10.1126/science.aai7488

Source: science.sciencemag.org

Mutualistic networks have been shown to involve complex patterns of interactions among animal and plant species, including a widespread presence of nestedness. The nested structure of these webs seems to be positively correlated with higher diversity and resilience. Moreover, these webs exhibit marked measurable structural patterns, including broad distributions of connectivity, strongly asymmetrical interactions and hierarchical organization. Hierarchical organization is an especially interesting property, since it is positively correlated with biodiversity and network resilience, thus suggesting potential selection processes favouring the observed web organization. However, here we show that all these structural quantitative patterns—and nestedness in particular—can be properly explained by means of a very simple dynamical model of speciation and divergence with no selection-driven coevolution of traits. The agreement between observed and modelled networks suggests that the patterns displayed by real mutualistic webs might actually represent evolutionary spandrels.

 

The architecture of mutualistic networks as an evolutionary spandrel
Sergi Valverde, Jordi Piñero, Bernat Corominas-Murtra, Jose Montoya, Lucas Joppa & Ricard Solé
Nature Ecology & Evolution (2017)
doi:10.1038/s41559-017-0383-4

Source: www.nature.com

Can live music events generate complex contagion in music streaming? This paper finds evidence in the affirmative, but only for the most popular artists. We generate a novel dataset from Last.fm, a music tracking website, to analyse the listenership history of 1.3 million users over a two-month time horizon. We use daily play counts along with event attendance data to run a regression discontinuity analysis in order to show the causal impact of concert attendance on music listenership among attendees and their friends network. First, we show that attending a music artist’s live concert increases that artist’s listenership among the attendees of the concert by approximately 1 song per day per attendee (p-value<0.001). Moreover, we show that this effect is contagious and can spread to users who did not attend the event. However, the extent of contagion depends on the type of artist. We only observe contagious increases in listenership for well-established, popular artists (.06 more daily plays per friend of an attendee [p<0.001]), while the effect is absent for emerging stars. We also show that the contagion effect size increases monotonically with the number of friends who have attended the live event.

 

Social Complex Contagion in Music Listenership: A Natural Experiment with 1.3 Million Participants
John Ternovski, Taha Yasseri

Source: arxiv.org