Recommender systems have been playing a central role in a variety of domains. When restricted to the application for citizens, a large body of literature could be found on Point-Of-Interest (POI) recommendation, tourist location recommendation, and orthogonally, spatio-temporal context-aware recommendation.
However, existing literature primarily focuses on a single domain (e.g. POI recommendation) from an algorithmic perspective, without considering from a personal perspective the closely related citizen activities as a continuous experience. Citizen activities in daily life, including transportation, working, sports, entertainment, shopping, etc., while being diverse, are closely related to each other. Together they describe a continuous experience for a citizen living in an urban environment. This calls for scientific investigation on the relationships among citizens' daily activities, in order to better understand their behaviors.
More importantly, the target user of conventional recommender systems are individuals, while citizens' recommender systems serve the society as a whole. To optimise the effectiveness of recommendations to the society, citizens' recommender systems require deep understanding of citizen-environment and citizen-citizen interactions. For example, to optimally recommend driving routes to a community of citizens, the system should be able to understand how the effectiveness of recommendations is influenced by the environment (e.g. road condition) and how the recommendations to different citizens affect each other (e.g. to avoid traffic congestion). The latter example shows the need for either algorithmic design for the recommendations to benefit the society as a whole, or incentive mechanisms to balance personal and societal interests.
The success of citizens' recommender systems heavily depends on the amount and quality of citizens' behavioural data, which were used to be scarce. The scarcity, on the one hand, can be compensated by engaging citizens to actively contribute their behavioural data to the system, i.e. citizen crowdsourcing via effective incentive schemes. On the other hand, the emergence of social data, i.e. data generated by people during their societal activities, available through new sources (e.g. social media, mobile phones, sensor networks), brings both opportunities and challenges to the development of citizens' recommender systems. Such data, when well-integrated, contain a multitude of dimensions, such as the targeted urban population, the purpose of use, the spatio-temporal context, etc.. Thus they describe comprehensively citizens' behaviors and their relationships with the environment, providing opportunities for recommendation based techniques to enhance citizens' living experiences. At the same time, social data are characterized by an intrinsic diversity, manifested through each of the relevant dimensions. It remains an open question how to leverage such diversity for optimally enhancing citizens' living experiences.
Recent studies have shown that recommender systems can actively change citizen mobility patterns, reducing traffic congestion and improve urban mobility. By exploiting social data and addressing the challenges, recommender systems have the potential to largely expand their impact and play an important role in today's society, in improving citizens' living experiences and the effectiveness of environmental use.
The topics of interest include but are not limited to:
- Requirements definition, design and implementation for citizen recommendation
- Collection, integration, exploration of social data for citizen recommendation
- Citizen user modeling and behavioral analysis
- Mining social data, social urban data for citizen recommendation
- Crowdsourcing for citizen recommendation
- Group recommendation in citizens' recommender systems
- Algorithms for citizen recommendation
- Incentivazation in citizen recommendation
- Spatio-temporal context in citizen recommendation
- Revisiting of POI recommendation in urban environment
- Cross-domain recommendation for citizens' continuous living experiences
- Citizen recommendation for smart urban environment
- Design, implementation of citizen knowledge base, and knowledge transfer to citizen recommendation
- User interface for citizen recommendation
- Ethical, cultural issues related to citizen recommendation
- Privacy and policy in citizen recommendation
The workshop accepts
- regular research papers (4 to 6 pages), and
- application/demonstration papers (2 pages)
- position papers (2 pages)
in ACM conference format (references are counted in the page limit). All of the submissions should be submitted via EasyChair system: https://easychair.org/conferences/?conf=citrec2017
After the workshop, the participants will come together to write a position paper about the potential and challenges of citizen recommendation.
The accepted papers and the collectively written position paper will be published in the ICPS series of ACM Digital Library (i.e., the choice of RecSys Challenge and DLRS workshops in 2016).