A View Behind the Scene: Data Structures and Software Architecture of a VAA Constantinos Djouvas and Nicolas Tsapatsoulis Dept. of Communication and Internet Studies Cyprus University of Technology {costas.tziouvas, inicolas.tsapatsoulis}@cut.ac.cy Abstract—In today’s rapidly evolving and growing online community many different applications are proposed and im- plemented. One category of such applications that drew high attention during the last few years are the so-called Voting Advice Applications (VAAs). VAAs are online systems used during elec- tions that allow voters to create a political profile; the comparison of this profile with the profiles of political parties and candidates provides the voter with an estimation of his/her proximity to parties and candidates. In this paper, the data structure and the software architecture used for implementing a VAA platform along with its technical requirements are presented. Furthermore, a novel approach used for supporting multi-lingual content is described. I. I NTRODUCTION The impact of the World Wide Web on politics around the world created a new era in the political landscape. Towards this direction, different online applications are proposed and developed. One such category of applications is the so-called Voting Advice Applications (VAAs). Voting Advice Appli- cations (VAAs) are online tools that allow voters to create their political profile in terms of their views on some policy issues [1]. By comparing a voter’s profile with the positions of political parties or candidates and utilizing different statistical methods, a VAA generates a number of results presenting the voter’s proximity to parties / candidates. Voters can utilize these results in order to see where they stand compared to parties or candidates. In order to provide the above-mentioned statistical results, a VAA must collect and maintain different types of information. These include: (1) the policy issues (i.e., a set of questions) based on which the voters and parties / candidates will gen- erate their profile, (2) the generated profiles of the voters and of the parties / candidates, and (3) possibly a set of optional questions that facilitate the acquisition of related information on a voter, e.g. age, gender, previous vote decisions, etc. Advances in VAA research resulted in a new type of VAAs referred to as Social VAAs (SVAAs) [4]. SVAAs support VAA friends, allowing them to share and compare their results. Thus, SVAAs, in addition to the above-mentioned information, collect all friends related data. Finally, a VAA must utilize the appropriate technologies for interacting with voters, collecting all the necessary informa- tion, and elegantly presenting the generated statistical results. Despite the fact that various experiments with such tools have been reported, to the best of our knowledge, there are no reports on their data and software structure and architecture. In this paper, we present the technological aspect of a VAA. More precisely, we give a detail description of the required technologies and data structures utilized for building a VAA. II. BACKGROUND VAAs are online tools that produce a match between the political positions of voters and parties / candidates [6]. For a VAA to produce such results, a number of planning and configuration steps are required. The first is the compilation of a questionnaire consisting of policy questions grouped into different topics relevant to the voters (i.e., economy, foreign policy, etc.). The process of selecting the appropriate questions to form the questionnaire is a vital step for the success of a VAA. This is because questions must bring out some the most pertinent issues in an election campaign and even raise awareness for political issues and policy decisions on some- times neglected topics [5]. At the same time, questions must polarize parties / candidates, showcasing their differences. In a different case, all parties / candidates will appear to have similar position on the under question policy issues. The second step, which usually runs in conjunction and influences the first step, is parties / candidates encoding. A team of academics and researchers utilizes a number of sources (i.e., policy statements, speeches, independent experts’ opinions, etc) to best estimate the answer a party would have given to the policy questions. A number of academics and researchers from different institutions may collaborate in order to increase credibility. For example, in the case of the VAA for the parliamentary elections in Greece 1 , researchers and academics from five different institutions and countries collaborated in order to create an independent questionnaire and encode the parties’ / candidates’ positions. Another approach for party / candidate encoding is to allow parties / candidates to answer the questionnaire themselves. However, this approach allows parties / candidates to manip- ulate the encoding, providing the most favorite answers and not answers based on their ideological positions and long term statements. The final step is the upload of the questionnaire and the encodings to a VAA platform. This will allow users to fill in the questionnaire and observe the generated results. The way 1 http://www.choose4greece.org