Search

In this keynote, I will present some results we have obtained using location data from mobile phones interacting with information (news) websites, mobile apps like Pokemon GO and Twitter, and large physical spaces like shopping malls. I will draw some conclusions and generally discuss about the properties of mobile phone data for location-based research to finally close with some remarks about privacy and data security.

Since 2016, through an association between Telefonica R&D and the Institute of Data Science in Chile, a group of researchers and myself have been working with trillions of digital traces left behind when people use their mobile phones. All of this work has been done under the general umbrella term of "data science for social good", and we have worked on anything from population displacement after external events like earthquakes, how people started using public spaces after the introduction of a popular mobile game, to actual social inclusion of people of different socio-economic backgrounds mixing in shopping malls or reading certain kinds of news, or patterns arising from gendered data sets. We will show how data in the private sector made us learn important social lessons such as how parks can become more secure when people went out to play Pokemon Go, how certain malls are hubs of social inclusion, how gender segregates the city and how different demographics keep themselves in their own informational filter bubble. However, even after all this benefits, the relationship with industry has never been fluid, and involves a lot of small and not so small compromises and "battles". In this talk, I will present a technical history of the work we've done with X/CDRs for social good including practical aspects of accessing and sharing data, the balance of research and industrial innovation, and issues of transactions costs while still providing value for the company itself, government, the university and society. I will also recount experiences about what it meant for a company like Telefonica and a research university like us to travel together in a very interesting context of huge data, incredible insights, privacy considerations, money, corporate interests, university expectations, and data-driven discovery.

The always increasing mobile connectivity affects every aspect of our daily lives, including how and when we keep ourselves informed and consult news media. By studying a DPI (deep packet inspection) dataset, provided by one of the major Chilean telecommunication companies, we investigate how different cohorts of the population of Santiago De Chile consume news media content through their smartphones. We find that some socio-demographic attributes are highly associated to specific news media consumption patterns. In particular, education and age play a significant role in shaping the consumers behaviour even in the digital context, in agreement with a large body of literature on off-line media distribution channels.

News consumers expect news outlets to be objective and balanced in their reports of events and opinions. However, there is a growing body of evidence of bias in the media caused by underlying political and socio-economic viewpoints. Previous studies have tried to classify the partiality of the media, but there is little work on quantifying it, and less still on the nature of this partiality. The vast amount of content published in social media enables us to quantify the inclination of the press to pre-defined sides of the socio-political spectrum. To describe such tendencies, we use tweets to automatically compute a news outlet's political and socio-economic orientation. Results show that the media have a measurable bias, and illustrate this by showing the favoritism of Chilean media for the ruling political parties in the country. This favoritism becomes clearer as we empirically observe a shift in the position of the mass media when there is a change in government. Even though relative differences in bias between news outlets can be observed, public awareness of the bias of the media landscape as a whole appears to be limited by the political space defined by the news that we receive as a population. We found that the nature of the bias is reflected in the vocabulary used and the entities mentioned by different news outlets. A survey conducted among news consumers confirms that media bias has an impact on the coverage of controversial topics and that this is perceivable by the general audience. Having a more accurate method to measure and characterize media bias will help readers position outlets in the socio-economic landscape, even when a (sometimes opposite) self-declared position is stated. This will empower readers to better reflect on the content provided by their news outlets of choice.

This article reports a case study of the iterative design and evaluation of a natural language-driven assistive technology, iGraph -Lite, providing people who are blind access to line graphs. Two laboratory-based usability studies involving blind and sighted people are presented with a discussion of the ensuing implementation of changes. Blind participants were found to adopt different graph interrogation strategies than sighted participants. A small field study is then reported in which a blind user who works with graphs took part to determine the degree to which the iGraph -Lite commands would meet the needs of blind graph experts. The final study invited sighted graph experts and novices to visually inspect and explain a set of line graphs comparable to those used in the usability studies. It aimed to highlight the concepts and the range of words sighted people use, to ascertain the appropriateness of the iGraph -Lite lexicon. A set of preliminary guidelines is presented.

Mathematics accessibility is an important topic for inclusive education. In this paper, we make Wikipedia's repository of mathematical formulas accessible by providing a natural language description of its more than 420,000 formulas using a well-researched sub-language. We also contribute by targeting Spanish speakers, for whom assistive technologies, particularly domain-specific technologies like the one described here, are scarce. Our focus on the semantics of formulas rather than their visual appearance allowed us to generate verbalizations with a precision of approximately 80% of understandable descriptions, as shown in an evaluation with sighted users.

Although pictorial renditions of statistical data are ubiquitous, few techniques and standards exist to exchange, search and query these graphical representations. We present several improvements to human-graph interaction including i) a new approach to manage statistical graph knowledge by semantic annotation of graphs that bridges the gap between Web 2.0 social tagging and formal, logic-based approaches, ii) knowledge management and discovery across a non-trivial graph knowledge base and iii) sophisticated question answering that requires background knowledge using Semantic Web technology.

Designing graphs and charts visually by means of graphing applications such as OpenOffice or MS Excel is extremely efficient and cost-effective. However, one of the drawbacks of such approach is that graphs are sometimes involuntarily made less accessible by, for instance, using a text box as title. In this paper we evaluate a corpus of 120 ecologically-valid statistical graphs for accessibility problems, discuss possible algorithms to solve these problems and finally propose the OM (Object Model) Principle, which states that any digital object is made more accessible by simply using the application's model for that object: for instance, the TITLE field for the title text.

Numerical information is often presented in graphs to take advantage of the human ability to quickly find visual patterns. Unfortunately, this medium is problematic for people who are blind or otherwise visually-impaired. To provide accessibility to graphs published in The Daily (Statistics Canada's main dissemination venue), we have developed iGraph, a system that provides short verbal descriptions of the information depicted in graphs and a way of also interacting with graphical information.

Statistical graphs are ubiquitous mechanisms for data visualization such that most, if not all, enterprises communicate information through them. However, many graphs are stored as unstructured images or proprietary binary objects, making them difficult to work with beyond the reports in which they are embedded. While graphs can be mapped to more common XML representations, these lack expressive semantics to discover new knowledge about them or to answer queries at various levels of granularity. This paper describes an OWL ontology that facilitates the representation, exchange, reasoning and query answering of statistical graph data. We illustrate the advantages of using an ontological approach to discover and query about time-series statistical graphs.

Much numerical information is visualized in graphs. However, this is a medium that is problematic for people with visual impairments. We have developed a system called iGraph which provides short verbal descriptions of the information usually depicted in graphs. This system was used as a preliminary solution that was validated through a process of User Needs Analysis (UNA). This process provided some basic data on the needs of people with visual impairments in terms of the components and the language to be used for graph comprehension and also validated our initial approach. The UNA provided important directions for the further development of iGraph particularly in terms of interactive querying of graphs

Numerical information is often presented in graphs. However, this medium is problematic for certain audiences such as inexperienced graph readers, people with visual impairments and users of mobile technologies featuring small screens. We have developed a system called iGraph which provides short verbal descriptions with the information depicted in graphs and a way of interacting with it by means of dialogue in natural language. In this paper, we present the general architecture of the system and its representational and querying mechanisms, together with a glimpse of the natural language interface.

This paper outlines our project to help beginners learn to program by showing object visualisations driven by aspect- oriented programming, and presented as part of the Eclipse development platform. The aspect-oriented programming is part of the infrastructure we use to drive the visualisations that help students learn object-oriented programming. Aspect-oriented programming explicitly supports the kind of cross-cutting concerns that allows our system to drive visualisations that emphasise principles of object interaction. Our extensions to Eclipse allow us to provide this educational scaffolding to help learners, without altering the program, the programming language or the libraries.

The purpose of this paper is to discuss the output of a neural network based on a linguistic model for recognizing the levels of involvement of different verbal arguments, assuming the non-discreteness of thematic relations and their non-primitiveness in linguistic theory. The network's output called for hypothesizing that, contrary to the received view, there is no equal level of involvement in verbal arguments, one having to be always more invovolved than the others, even when having the same number of Proto-Agent and Proto-Patients contributing properties.