New publication: Where's everybody? Comparing the use of heatmaps to uncover cities' tacit social context in smartphones and pervasive displays.

We introduce HotCity, a city-wide social context crowdsourcing platform that utilises user's current location and geo-tagged social data (e.g., check-ins, "likes" and ratings) to autonomously obtain insight on a city's tacit social awareness (e.g., "when is best time and where to go out on a Saturday night?"). HotCity is available as a mobile application for Android and as an interactive application on pervasive large displays, showcasing a heatmap of social buzz. We present the results of an in-the-field evaluation with 30 volunteers, of which 27 are tourists of the mobile app, compare it to a previous evaluation of the pervasive display app and also present usage data of free use of the pervasive display app over 3 years in the city of Oulu, Finland. Our data demonstrate that HotCity can communicate effectively the city's current social buzz, without affecting digital maps' cartography information. Our empirical analysis highlights a change in tourists' foci when exploring the city using HotCity. We identify a transition from "individual [places]" to "good [areas]" and "people [choices]." Our contributions are threefold: a long-term deployment of a city-wide social context crowdsourcing platform; an in-the-field evaluation of HotCity on mobile devices and pervasive displays; and an evaluation of cities' tacit knowledge as social context as a denominator in city planning and for the development of future mobile social-aware applications.

Komninos, A., Besharat J., Ferreira D., Kostakos V., & Garofalakis J. (2017).  Where's everybody? Comparing the use of heatmaps to uncover cities' tacit social context in smartphones and pervasive displays.. Information Technology & Tourism, Springer. Online First, .. DOI:10.1007/s40558-017-0092-5
Download full text: 

New publication: Natural interaction with large map interfaces in VR

Location based services are a common application scenario in mobile and ubiquitous computing. A major issue with map applications in this domain is the limited size of the display, which makes interaction and visualization a difficult problem to solve. With the increasing popularity of VR and AR systems, an opportunity exists for map-based applications to overcome the limitation small display sizes, as the user’s information visualization space can extend to her entire surroundings. We present a preliminary investigation into how interaction with such very large display interfaces can take place, using a virtual reality headset as the sole input and interaction method.

Giannopoulos, I., Komninos A., & Garofalakis J. (2017).  Natural interaction with large map interfaces in VR. 21st ACM Panhellenic Conference on Informatics. Larisa, Greece, ACM. DOI:10.1145/3139367.3139424
Download full text: 

New publication: A lightweight algorithm for the emotional classification of crowdsourced venue reviews

Finding emotions in text is an area of research with wide- ranging applications. Analysis of sentiment in text can help determine the opinions and affective intent of writers, as well as their a itudes, evaluations and inclinations with respect to various topics. Previous work in sentiment analysis has been done on a variety of text genres, including product and movie reviews, news stories, editorials and opinion articles, or blogs. We describe a lightweight emotion annotation algorithm for identifying emotion category & intensity in reviews wri en by social media (Foursquare) users. e algorithm is evaluated against human subject performance and is found to compare favourably. is work opens up opportunities for solving the problem of helping user navigate through the plethora of venue reviews in mobile and desktop applications.

Kolokythas, P., Komninos A., & Garofalakis J. (2017).  A lightweight algorithm for the emotional classification of crowdsourced venue reviews. 21st ACM Panhellenic Conference on Informatics. Larisa, Greece, ACM. DOI:10.1145/3139367.3139422
Download full text: 

Pages