The paper Detecting air travel to survey passengers on a worldwide scale (pre-editing version), co-authored with Pierre Dillenbourg and Nicolas Nova, as been accepted for publication in the Journal of Location Based Services. It reports on a methodology for gathering mobility data at a world-wide scale, contrasting with traditional travel survey methods. The originality of this research work is to take into consideration the limitations of the technological settings as well as the complexity of human and technological environments as the source of the design solution.

Abstract. Market research in the transportation sector is often based on traditional surveys, such as travel diaries, which have well documented shortcomings and biases. The advent of mobile and wireless technologies enables new methods of investigation of passengers behaviour that can eventually provide original insights into mobility studies. Because these technologies can capture travellers’ experience in context and real time, they pave the road for new surveys methods. In this paper, we demonstrate that mobile phones can recognize air travel with a light algorithm that scans their connectivity to cellular networks. The originality of our method is that it does not rely on any GPS-like location information and runs on a large variety of mobile phones. It detects flights on a worldwide scale and asks travellers to report on their travel experiences as they occur, eliminating the recall bias of traditional solutions. Once the system detects a journey, it triggers a flight satisfaction questionnaire that sends answers to a centralized server. This approach respects the traveller’s privacy and proved a 97% success rate in detecting flights in a 12-months study involving 6 travellers who boarded on 76 planes.

Keywords: Sensing and activity recognition, mobility detection, transportation study

Unlike the traditional ways to capture travel information, our approach relies on the mobile phone to generate “automatic passive” GSM fingerprints and trigger an in-situ questionnaire. It is an hybrid solution of implicit motion detection with the air traveller’s consent and explicit disclosure of the travel experience. The motion detection is based on an algorithm that analyses the sequences of GSM network Location Area Identity. The figure below shows 3 examples: In flight, the mobile phone roams from one country code to another. If our software does not retrieve any LAI within 30 minutes, it detects an air travel. In a train, the mobile phone moves within different network providers and area codes. No survey appears if the disconnection periods do not exceed 30 minutes. Similar scenario takes place for a car that moves within different location areas.

Why do I blog this: Work conducted a couple of years ago as a fruit of the research developed for CatchBob! at the demand of a client of Simpliquity. Unlike many designs that consider practical constraints as detrimental to the elegance of technological solutions, we instead viewed them as opportunities to rethink solutions which eventually have to change over time.

It is entitled “Aspects of implicit and explicit human interactions with ubiquitous geographic information“. In this dissertation, I cover some implications of the presence of ubiquitous technologies that afford people new flexibility in conducting their daily activities and simultaneously provide the means to study human activities in time and space. I employ seven of my first-authored papers to describe the following aspects of people interactions with these technologies to access and generate ubiquitous geoinformation:

• Sources and perception of spatial uncertainty (CatchBob!)
• Appropriation of ubiquitous geoinformation (Taxi drivers in Barcelona)
• Implicit human interaction with wireless infrastructure as source of travel detection (Travel survey, no public content yet)
• User-generated ubiquitous geoinformatin as evidence of tourist dynamics (Tracing the visitor’s eye in Florence and Rome)
• Digital footprints as evidences of urban attractiveness (New York City Waterfalls)

At the start of this journey to a PhD, 3.5 years ago, I remember my wish to learn to fit to the mold of academia. Instead, I quite naturally emancipated from that thought, with inspirations from Nicolas or Julian. While keeping true to the purpose of science, I certainly intended to explore beyond the bounding box of one domain and its methods. Reading Järvinen Pertti’s book gave me that necessary holistic view on research methods. The exposure to new approaches and different research cultures and the consideration of real-life issues certainly contributed to the unusual aspect of my thesis. In fact, this thesis is the fruit of this exploration. I am convinced that the interdisciplinary knowledge and skills developed in this journey will enable me to adapt to a constantly changing world. Indeed, my work does not claim in-depth findings acquired through the application of one methodology. Rather, It stresses that varied approaches applied to varied problems strengthen both the methodological approach and the support for the more factual knowledge evidenced. To reach this contribution, I explored the creation of footbridges between domains that do not have the tradition to cross-communicate. I take pride to have developed both factual and designl knowledge from field works in collaboration with cognitive scientists, social scientists, software engineers, network engineers, computer scientists, industrial designers, and architects. The warm response of the members of my thesis jury constituted by experts from the domains of geography, anthropology, data mining, computer science, spatial information science and transportation research is certainly a proof of the attractiveness of my work in a wide context. It also confirms my wish, shared with my PhD advisor, to produce a scientifically honest piece of work that is accessible to people on the edges of academia.

In my next journey, I will further nurture my research qualities, on that edges, at the friction area between the technical, the human and the urban. But this time, I will practice freed from the overheads of academia (more on that later).

Relation to my thesis: Almost the end of the process… let’s prepare the defense! A collateral outcome of this journey, I now master a fourth language, Spanish.

Related to *sigh* DEA Thesis Submitted *sig* and Framing my PhD Dissertation

To complete my PhD dissertation that will be based on a series of papers, I am writing a journal version of my study presented at last year’s AAG: The co-evolution of taxi drivers and their in-car navigation systems. Besides improving the details that describe the research method, I need to rephrase the elements of co-evolution I aimed at capturing, re-categorize the findings and work on a discussion section that compares the findings with related works and link them to implications consideration for design. A revamp I dissect below.

Motivations
The extensive distribution of satellite navigation systems represents the first massive adoption of location-aware system by the public. However, the market success also translates into documented poor integration of the technology into the practice of driving (e.g. distractions that create accidents, frustrations due to the uneven quality of the geoinformation). From this observation, we hypothesize that users of satellite navigation systems overcome these types of problems by both adapting to their systems and adapting their systems to their needs. First we wanted to report on evidences of this co-evolution (theorized in Orlikowski, 1992a; O’Day, Bobrow, Shirley, 1996, Ackerman, 2000) and then describe the detected implications (e.g. (de-)skilling factors on the knowledge of the city and social relationships, modification of the wayfinding practice, new relations to the environment). An ethnographical approach provides an appropriate analytic angle to examine user practices in context and situated activities (need stronger arguments here!). We argue that our observations and interpretations can be valuable to designers of next generation location-aware systems particularly to understand how a) users create over time unanticipated use to better fit their need, if the system is capable of only partial satisfaction; b) a location-aware system effects the interaction of the user with their environment (social, physical, informational); c) the implication of the adoption of location-aware system on potential for future design. An example in Hutchins cognition in the wild is the slow evolution of navigational tools. For example, the Mercator projection for maps simplifies navigation calculations, and its creation was an act of tool adoption and co-evolution.

Phases of co-evolution
We analyzed this co-evolution for three different phases

1. Acquisition: why and how this new technology gets integrated into other artifacts and how, over time, it impacts the use of these artifacts;
2. Expectation gap: how the reasons to acquire are matched in practices and where are the expectation gaps;
3. Evolution of the appropriation: despite the gaps, how does the practices evolves in relation a) to the awareness and reactions to the limitations and imperfections of the system; b) to driver’s knowledge of the environment; c) to the access to geoinformation both from the system and other artifacts.

Approach, method and participants
So the market success of makes relevant phenomenon to study how people integrate pervasive geoinformation into their practices in the context of their work. What was even more appealing was to perform a systematic study of one type of early adopters, the taxi drivers in Barcelona (available in a large number of the 10,400 licensed cars), who have massively adopted the technology the last few years. This context is precisely relevant to the research on location-based systems because it provides evidences, outsides of the laboratories of their impact and implications, we believe necessary to feed the design process. The study of in-car navigation systems are typically interested in usability issues with an experimental angle. Rather than focusing on the device, we are interested on the practice with mobile and location-aware devices (see Understanding mobility contexts), on the losses and opportunities, and considering a full-circle perspective (Chrisman, 2005)

In our ethnographically-inspired study, we collected empirical evidences from 12 taxi drivers (11 males, 1 female) with a working experience from 2 to 20 years. In a first phase, we conducted semi-structered interviews at the Barcelona airport parking lot dedicated to taxis. This site provided a good setting to get in touch with taxi drivers, because they often must wait there for more than 30 minutes. During this period, they fill the time discussing, playing (e.g. chess, scrabble, golf), cleaning their care, taking a refreshment and eating. The navigation system had been purchased and used by the informants for a least 6 months. Interviews alone did not suffice, as the informants necessarily selective in what they describe or think is important to discuss. Therefore, they complement additional situational insights from observation of driver work settings (e.g. artifacts used) and behavior (e.g. engagement with geoinformation) while driving. Acting is customers, we requested rides from and to a railway station, a major hotel and an address on one of the many very narrow streets of the Gracia neighborhood. Each informant participated in one session, lasting approximately in from 45 minutes to 1 hour, divided in half with situational insights then complementary semi-structured interviews in the other half.

Despite the small sample of taxi drivers, we were able to collect rich data consisting of photos of work settings (to record the eco-system of artifacts), videos of taxi rides (to capture key moments of access to geoinformation) and written field notes from the semi-structured interviews (analyzed and compacted after each session). In our analysis consisted in a careful reading of the different materials and highlighting parts that were related to the phases of the co-evolution we wanted to focus on. We organized the relevant parts into common themes, and coded the documents using the emerging themes. The sessions were conducted in Spanish and we provide the closest translation for this paper. We summarize the findings extracted from these themes in the following section.

After a red light, in proximity to destination, a taxi driver is about to close his “Guia” after checking the location of a street number. The navigation system is in passive mode.

Findings

Addition to an existing ecosystem of artifacts
The navigation system completes an arrangement of artifacts proving layers of geoinformation to support the job. With our observations, we were able to categorize detected the real-time information sources with the sat-nav system, dispatched radio, electronic booking system, mobile phone and radio. Another category of artifacts provide more static information of the city; the newspapers to keep the knowledge of events and activities in the city up to day, scribbled list of “unofficial” points of interests requested by customers such as the strip clubs, and above all the street directory and map of Barcelona “Guia Urbana de Barcelona” generally kept at reach on the passenger seat, above the dashboard or the interior storage compartment. (Need a table to describe the artifact, its purpose, context of use and the evolution of this ecosystem.

Serenity over efficiency
In the past, taxi drivers face moments of uncertainty when a customers take them to a village or unfamiliar areas of Barcelona, such as the multiple suburban business areas that emerged from the ground in the recent years. Nowadays, there is a consensus that a satellite navigations system is an unbeatable tool when it come to reaching a specific destination in a village and leaving it and it is a strong reason for purchase. But it is not really to be more productive and hence earn more, rather than feeling more serene in appropriately doing their job. A tool not to make more money, nor to improve the efficacy, but to tranquilize, ”I can go everywhere and relax” said informant X and “the fear of getting lost with a customer I felt in my stomach now disappeared” (informant Y). Other drivers mentioned that it does not only tranquilize them but also their customers ”they know I cannot cheat them with it”, “it reassures them that I go to the proper destination”. This feeling of relying on a “companion” for critical situations reveals particularly when informants referrer to it as a “he” (and their car as a “she”). That does not prevent them in keeping it most of the time in the glove compartment and retrieving it only the times it is necessary.

The phases of wayfinding
The wayfinding practice takes place in 2 main phases. For the “to go” phase in which the driver takes the proper direction and the “to arrive” phase in which the precise drop off location (the number of the street, a monument, a corner) necessitate different geoinformation at a different granularity. […] For a ride to a rather unknown destination, drivers seem to accesses the geoinformation as in a “funnel”. First, there is a tendency to quickly retrieve in the paper-based street directory the area of the destination point. If the paper map does not provide enough information such as reference points to precisely arrive at the area, the drivers we observed engage with the navigation system. No detailed information are requested at that moment of the ride. It is only when approaching the destination area, that the exact address in the navigation system is entered. Informant X explained that he engages with the system at that moment to avoid the often misleading information on the path to take. Indeed, a taxi driver applies several paths depending on the time of the day and circumstances (e.g. traffic, weather conditions, preference of passenger).

An altered learning process
In the past, the driver would open the “Guia” and browse the index of streets and points of interest to get access to a map of the area comprised by the destination. The book was one way to learn the city by doing. Another big part of the knowledge drivers acquire come from the customers themselves, as they communicate the tricks and secrets that official books and commercial systems do not disclose. As the experience drivers contain most of the knowledge of the city in memory they rarely return to the Guia. In fact, they prefer using their sat-nav and open the Guia only as fail-over (e.g. when the system does not contain a street number or lacks of information). Even after 20 years of experience, one drivers still perceived reading the Guia as a demanding task. Compare to that, “hitting the screen of his Tom-Tom is a pure pleasure“. Experienced drivers only rely in the system (sometimes keeping it in the glove compartment - or only use it to keep track of their speed) when driving to unfamiliar destinations, which after 20 years only happens rarely. The system is not source of learning but rather a punctual saver.

The integration of the satellite navigation system altered the learning process of the city from three main disruptions: the necessity to assess the quality of the geoinformation, the imputation of some social interactions and the ability to temporarily disengage with the environment:

- Assessing the quality of geoinformation
Taxi drivers with a short job experience express mixed feelings about their experience with their sat-nav ”It is like my mobile phone, sometimes it does not work well”, “it is a potential problem” in terms of quality of services and precision of the information. Their demand of accurate information quickly reached the limits of the systems and the inaccuracy of the system directly impacted the quality of their service. Because they had to deeply rely on it in the beginning, they have learned when not to rely on it. For instance, informant X could name me the places where he experienced the navigation information to be absolutely irrelevant (e.g. get access major squares). But that does not make them learn the city. In consequence, they embrace the use of the “Guia” which is “more accurate and complete” for their job, particularly when it comes to providing a detailed index of streets, points of interest and city-related information.

- New engagement with the environment
As drivers cannot completely rely on their navigation system, they have to form their knowledge of the city to improve quality of their service and relax. One mean translated into leaving their navigation system system on passive mode to learn the street names or identify the present of fixed speed measurements radars as they appear on the screen. Keeping the system on passive mode, also enables the driver to disengage with the environment for a while knowing that the screen will give a proper position. In the past, drivers would count the number of streets to cross before making a turn.

- Social amputation
A major part of about learning about the city comes from the interaction with the customers. It is still a tradition for a taxi driver to ask “how do you want to go there?” and rely on the direction given by the customers to avoid any complains and learn the tricks to navigate. The presence of a navigation system changes this social configuration and reduces the opportunities for this learning process to take place. This technological presence not only affects the social interactions not only inside the car, but also with remote friends and colleagues. Informant X was relieved that he did not have to use his radio to ask colleagues about particular location information and directions (even though many taxi drivers are social animals). Similarly, taxi drivers have now less a tendency to ask for directions to locals (e.g. in remote villages) or other knowledgeable people of the city (e.g. truck drivers at red lights in industrial areas).
Discussion

- De-skilling?
Previous studies on navigation systems have highlighted that introduction of navigation system has shown evidences of de-skilling effect on orientation and navigation (Leshed et al., 2008, Elaluf-Calderwood and Sørensen, 2008, Aporta and Higgs, 2008). For instance “The more the driver relies on the system to locate jobs, the less he or she relies on their in-depth knowledge of where they need to position themselves to maximise income” ( Elaluf-Calderwood and Sørensen, 2008). This is also supported by the perception of the experienced generation of taxi drivers we got to observe and interview. Informant X explained that “the newcomers who use a navigation system do not gain knowledge of the city, because they follow the recommendations and stop to think“. Our analysis of of the less knowledgeable taxi drivers showed somehow has not revealed negative affects on their acquisition of knowledge that support their skills to perform well their job. We notices that they are eager to learn the city and the imperfection of their systems forced them to learn using traditional sources such as the “Guia”. These drivers have a tendency to keep their system passive mode, not only to detach with the orientation process but also to learn the environment in familiar areas such as keeping track of the street names and their sequences.

This need by less experience drivers to access multiple sources of information from an ecosystem of artifacts slightly breaks the myth that sat-nav system changes a “skilled” job unto a unskilled one; in other words anyone with a GPS could do their job. A prove as that the system gets to be used less and less over time, sometimes finishing in the glove compartments or even the trunk. However, the presence of the technology altered the social practices that were at the source of learning the “unofficial” city from the knowledgeable customers. In that sense taxi drivers may slowly change social practice to adapt, but we have no evidences of that. What is clear is that the navigation system does not provide these kind of information.

- Information quality, system limitations and co-evolution
The fluctuating quality of the geoinformation as sources of co-evolution of the taxi drivers and their in-car navigation system. An ecosystem of artifacts are set to complement the limitations of information. Indeed, taxi drivers learned not to completely rely on their sat-nav because they do not have the knowledge to assess the quality of the information. Therefore, besides learning to interpret the system and its inaccuracies, they have a tendency to open their “Guia”. Informant X admitted that when a conflict emerges between his sat-nav and his intuition he starts to “improvise” and depending on the circumstances he would either switch it off or ignore it for a while, a pure example of Suchman’s situated action theory (locations and opportunity determine the action): “Idiosyncrasy, improvisation and knowledge are all useful tools when choices between planned and situation acts are complex” (Suchman, 1987)”
- Considerations for design

* Moving beyond technological determinism

* People may use systems in ways unanticipated by the designers (more evidences for the discourse of Ackerman). For instance, one taxi drivers used the speed displayed in the navigation system rather than fixing his broken speedometer.
* System designers should assume that people will try to tailor their use of in-car navigation systems; a system that complements rather than replace an ecosystem of artifacts, sources of the knowledge of the city. It does not replace, because a large part of the knowledge has no informational model or digital sources.
* Support and not reduce social interaction with customers as opportunities to learn: often a taxi ride is an opportunity to engage in social interactions. A navigation system is primarily designed for a specific interaction with the driver without engaging with other people in the car, secluding the customer.

* Highlight the quality of the geoinfomation to help users to match their knowledge to assess the quality of geoinformation. This is particularly importance with the emergence of real-time traffic information. So the design should reveal the quality and timeless of the information (seamful design)

Relation to my thesis: Using this blog to reformulate a paper for journal submission (that should be my last journal paper) in re-aligning the thoughts free from the academic paper writing overheads

A paper by senseable’s colleague Bernd Resch has received the Best Paper Award at the IEEE International Conference on Advanced Geographic Information Systems & Web Services for his paper:

Resch, B., Mittlboeck, M., Girardin, F., and Britter, R. (2009). Real-time geo-awareness - sensor data integration for environmental monitoring in the city. In IEEE International Conference on Advanced Geographic Information Systems Web Services. Cancun, Mexico.

Bernd has been pursuing the urban sensor data interoperability holly grail, building methods to integrate heterogeneous data sources that both lack of interoperable interfaces and have deficient coordination due to monolithic and closed data infrastructures. He has been developing the kind “Sensor Data Integration” system based on open sensing infrastructure making extensive use of open (geospatial) standards.

Congrats Bernd!

Relation to my thesis: I humbly contributed to a section on the potential implications of such a system on raising people environmental awareness.

The International Workshop on Urban, Community, and Social Applications of Networked Sensing Systems took place this Fall with a few contributions in relations to my research projects including on detecting modes of transportation: MobileSense - Sensing Modes of Transportation in Studies of the Built Environment and on using new urban actors as proxies to uncover and measure urban dynamics: Measuring the Pulse of the City through Shared Bicycle Programs.

I was at the OECD headquarters in Paris to attend the The 9th International Forum on Tourism Statistics. The conference covered the main themes related to the collection and analysis of quantitative tourism data, from the classic issues on visitors survey (e.g. border surveys, hotel surveys) to the coordination among countries through Tourism Satellite Accounts (TSA) to produce more advanced tourism statistics. These methods need to keep on adapting to the constant evolution of tourism and tourists behaviors. For instance, the current demographic evolution with western countries getting older produces new tourism services and reduces seasonalities. In reaction there is progress in the automatic collection of data to deliver short-term estimates in demand and supply data. However, the unique use of these traditional methods raise the risk of missing emerging phenomena and changes in tourist behavior such as the raise of trips with shorter stays or concerns in sustainability.

9th International Forum on Tourism Statistics at the OECD headquarters

In response to this evolution that requests new methodologies to capture tourist behaviors, I participated to the session dedicated on “new approaches to data collection from the supply side”.

As is the case for most or all fields of statistics where data is collected from the economic actors, there is an increasing awareness of the respondent burden. Statistical offices try to reduce or control this respondent burden by finding more efficient ways of collection the primary information (e.g. e-surveys) from for instance the accommodation establishments. To obtain more information on e.g. tourism flows or tourism activities, data producers can make use of registers or other administrative sources and combine them with the survey data.

Such sources can be general purpose sources (for instance to obtain more detailed geographical information such as urban versus rural tourism) or can be tourism-related sources (e.g. specific tourist taxes or eco-labels awarded to enterprises in the tourism sector).

This session wants to contribute to exchanging national or regional practices that are not entirely linked to country or region-specific information systems and therefore have relevance for a larger group of tourism statisticians. In addition, this session also aims at getting a better understanding of estimations methods for private or non-rental accommodation, to complete the statistics on collective accommodation. Indeed, a very significant share of tourists stays in private rented accommodation or free accommodation. To get a comprehensive insight into the total tourism flows into a country, estimates of these less visible types of accommodation are indispensable.

I presented the paper Uncovering the presence and movements of tourist from user-generated content (en français) in which in discuss the opportunity in using implicit (cellular network traffic) and explicit (georeferenced photos) digital footprints to uncover novel evidences and anecdotes on urban tourism (slides of my talk):

In recent years, the large deployment of mobile devices has led to a massive increase in the volume of records of where people have been and when they were there. The analysis of these spatio-temporal data can supply high-level human behavior information valuable to social scientists, urban planners and local authorities. This paper explores this hypothesis by reporting on new information revealed by this pervasive user-generated content. We present novel techniques, methods and tools we have been developing to explore the significance of these new types of data. In a case study of Rome, Italy, we showcase the ability to uncover the presence and movements of tourists from geo-referenced photos they explicitly make public, as well as from network data implicitly generated by users of mobile phones.

In a similar line of research Michel Houée (Odit France, France) presented an exploratory work on Estimating foreign visitors from highways toll payments. This methods used credit cards payments at French highway toll stations as digital footprints of visitors in transit. Thanks to the merging of estimates from all cooperating networks, it is possible to . The analysis of these data give produce a global picture of the visits for each toll station serving the French territory, in terms of volume as well as in terms of proportion of the total foreign visits. For instance it gives insights on the split by nationality of visits of a specific tourist location over a whole year, point out the differences of monthly seasonality profile among the nationalities, and when the cooperation is ancient like with the ASF network, it is possible to analyze visit trends according to the nationality. The first major validation consists in checking the accuracy of the estimation of the traffic of each nationality during the survey period as compared to the counts

Using the information at toll barriers along the networks at peak periods of visit of France, it is possible to identify the main routes followed and to observe how the traffic decreases progressively from the entrance gateways up to border points of exit from the French territory (figure below).

Germans in summer: a high proportion of transit towards Spain

A flow that I also capture when entering Spain from France with my Flickr data gathered for the south of France, Spain and Portugal (figure below, second map from the right).

Main flows of visiting photographers in the Iberian Peninsula in 2007

My work on georeferenced photos and cellular network traffic analysis in Rome was rather well perceived as a potential cost alternative approach that can improve the timeliness of the data collection. The limitations are known, the data do not cover exactly a territory and needs strong calibration with more traditional data with respect to privacy. However, they contain some evidences and anecdotes that add “energy” (i.e. a more qualitative angle, and reactivity) to analysis based on traditional quantitative data. This goes well with the utility to build tools to support decision-making but also support a discussion process.

Relation to my thesis: my works frames well with current concerns in the domain of tourism statistics does not want to miss emerging behaviors (influence of ICTs on tourism) and trends (use of ICTs to grasp these behaviors). Solutions based on the analysis of digital footprints can potentially capture what traditional methods can’t, in a cost and time effective manner.

The methods and tools I have developed at the MIT SENSEable City Lab to explore the significance of the new types of user-related spatiotemporal data have been published in the current issue of the IEEE Pervasive Magazine. In Digital Footprinting: Uncovering Tourists with User-Generated Content I describe how our approach helps uncover the presence and movements of tourists from cell phone network data and the georeferenced photos they generate.

Girardin, F., Calabrese, F., Dal Fiore, F. , Ratti, C., and Blat, J. (2008). Digital footprinting: Uncovering tourists with user-generated content. IEEE Pervasive Computing, 7(4):36–43.

Relation to my thesis: A follow-up work of the case study in Florence. In this paper I explain how different types of digital footprints visitors leave behind when visiting Rome, capture their presence at different places and moments. While, the analysis of georeference photos help define sightseeing areas, mapping the aggregated cellular network activity originated from foreign mobile phones reveal places of travel (train station) and lodging
(hotel areas).

Geovisualizations of the presence of (a) 932 tourist photographers and (b) 520,000 phone calls from foreign mobile phones in the Coliseum and Piazza della Repubblica area from September to November 2006. Both types of data cover the train station area in the proximity of the Piazza della Republica. The values in each cell are normalized.

Potential venues to submit this Summer’s research outcomes:

AAAI Spring Symposium 2009 - Human Behavior Modeling

This workshop will explore methods for creating models of individual and group behavior from data. Models include generative and discriminative statistical models, relational models, and social network models. Data includes low-level sensor data (GPS, RFID, accelerometers, physiological measures, etc.), video, speech, and text. Behaviors include high-level descriptions of purposeful and meaningful activity or abstractions of cognitive and affective states. These include activities of daily living (e.g., preparing a meal), interaction between small sets of individuals (e.g., having a conversation), mass behavior of groups (e.g. the flow of traffic in a city) and related internal user states.
Deadline: October 3, 2008

4th International Symposium on Location and Context Awareness
technical papers describing research results on systems, services, and applications to detect, interpret and use location and other contextual information. Context includes physiological, environmental and computational data whether sensed or inferred. In addition, context includes users’ activities, goals, abilities, preferences, interruptibility, affordances, and surroundings. With context, we can expect computers to deliver information, services, and entertainment in a way that maximizes convenience and minimizes intrusion. Developing awareness involves research in sensing, systems, machine learning, human computer interaction, and design.
Deadline: December 18, 2008

See previous list of CFP Special Issues, Workshops and Conferences

These past week, a couple of my follow-up works on the preliminary Tracing the Visitor’s Eye case study based on Florence have been accepted. I will be presenting the position paper “Assessing pervasive user-generated content to describe tourist dynamics” at the First International Workshop on Trends in Pervasive and Ubiquitous Geotechnology and Geoinformation on the need and methods to understand the users practices behind the generation of digital footprints and their relation to the urban space. In addition, the paper “Digital footprinting: uncovering the presence and movements of tourists from user-generated content” will be published in the IEEE Pervasive Computing Magazine, special issue on Pervasive User-Generated Content. It showcases the ability to uncover the presence and movements of tourists from geo-referenced photos they explicitly make public, as well as from network data implicitly generated by users of mobile phones.

Besides encouraging reviews on an “exciting area of research with lots of potential” some comments suggest some areas I should focus my current and future data analysis on:

Reveal the unexpected
- “it would be interesting to hear whether any of the findings were unusual/unexpected”
- “delve deeper into how “digital footprints” provide novel observations/results compared with traditional knowledge (expend to other nationalities)”
- “also focus on people taking photos in their own city (how are they different?)”

In consequence, I am exploring methods to reveal abnormal behaviors from time series, and to profile users from their use of the system and their mobility.

Space and user profiling
- “other automatic group categorization are possible (type of camera?), segmentation of the cellphone data as well”
- “what areas are more prone to be taken with a camera phone than a digital camera? and how does this evolves?”
- “hypothesis: tagged data is likely to be of fairly high quality (accurate and semantically meaningful)”
- “different types of tourists may geocode photos differently (American tourists are much less adept at geocoding location apart from thos most famous landmarks than Italian tourists”

Validation
- “conclusions are conjectures and not yet validated (they are obvious and not too controversial)”
- “methods of validation will likely have to be developed”
- “how to assess the impact of the biases in the data?”
- “in what ways does the passively collected data accurately reflect reality?”
- “additional investigation into the reasons for the discrepancies between datasets seems warranted. Do the demographics of photographers match the demotraphics of the mobile phone users? This additional analysis may help more convincingly explain the validation discrepencies and provide an avenue for correcting biases.”

In consequence, I am collecting hard data from traditional surveys that should represent the reality to compare to and calibrate to if they correlate.

Heterogeneity of the analysis
- “tie more together explicitly each analysis (presence and flow)”
- “are there other types of analysis (besides presence, flow, semantics)”

In consequence, I have to set an exhaustive list of what can be analyzed for each type of digital footprint and digital shadow. Then explain the scope of each type of analysis and how they relate to each other.
In consequence, I have extended the features that characterize users (e.g. type of camera, taken time and upload time difference, categories of tags used, …) in a correlation matrix. Use of proximity function and clustering techniques to define groups of users. Use of Characterization techniques as seen previously in Methods to Study Flickr Users Behaviors.

A couple of papers studying users behaviors in the context of online photo-sharing. Each has a specific angle on the data: social science (individual and group cooperation practices), data mining (who is looking at a photo) and visualization (understand the notion of place). I use them to inform the analysis of Flickr users practices to georeference and geotag their photos.

Prieur, C., Cardon, D., Beuscart, J.-S., Pissard, N., and Pons, P. (2008). The stength of weak cooperation: A case study on flickr. CoRR, abs/0802.2317.

Objective: Detail the concept of weak cooperation showing the great variety of uses (stockpiling, social media use, myspace like).
Data: 5M users, 150M photos, users, contacts, groups photos, comments, tags and favorites.
Methods: Distribution of flickr functionalities (contacts, comments, favorites); principal component analysis on correlation matrix to represent graphically the relations among nb photos, nb contacts (in/out), nb favorites (in/out); distribution of number of members and photos among Flickr groups, social graph with members of groups as nodes and proximity function between users from the similarity of the tags they use; define the social density of a group from the density of the social graph.

van Zwol, R. (2007). Flickr: Who is looking? In 2007 IEEE / WIC / ACM International Conference on Web Intelligence, pages 184–190. IEEE Computer Society.

Objective: Characterize user behaviors on temporal, social and spatial dimensions.
Data: 1.83M uploaded during 10 days and viewed over a period of 50 days.
Methods: Use a photo’s popularity to identify social networking and photo pooling. Distribution of photos views per bucket (a bucket groups the photos based on the number of views) 0-10%, 10-20%, 20-30%, 30-40%, 40-50% and > 50%. Same for the temporal dimension slices of time, and social dimension (comments, contacts, pools). Geographic distribution defined with two standard deviations (for the longitude and latitude) and a single value representing the geographic spreading of the photos views with the euclidian distance between the two standard deviations. This value is then applied to the slices.

Dykes, J., Purves, R., Edwardes, A. J., and Wood, J. (2008). Exploring volunteered geographic information to describe place: Visualization of the ’geograph british isles’ collection. In GIS Research UK (GISRUK 2008). Manchester Metropolitan University.

Objective: Understand the way language is used to describe landscape with geotagged photos
Data: 340,000 photographs with titles, comments and other metadata, georeference with at least 1km precision
Methods: Interactive and spatial treemaps of terms. Forcus on scene types (e.g. beach, village, mountain, hill) and scene type descriptors. Chi statistic maps of national trends for the selected combination of scene types. Spatial tag clouds to explore local variation in the terms.

Relation to my thesis:  Quantitative methods to understand user practices with online photo-sharing platforms in extensions to works such as Why We Tag: Motivations for Annotation in Mobile and Online Media