Aggregation of Constrained Crowd Opinions for Urban Planning

TL;DR

The paper tackles the problem of aggregating crowd opinions for urban planning under background infrastructure constraints. It introduces a generalized, unsupervised constrained judgment analysis framework that differentiates main and background constraints and demonstrates it on two case studies: point-based ATM location planning and line-based sewage planning. By employing centroid-based clustering with a Hausdorff distance and a pre-/post-processing constraint handling pipeline, the approach yields more reliable, feasible consensus while retaining a larger fraction of valuable opinions. The work provides a practical methodology for constraint-aware crowd judgments in smart city applications and suggests extensions to more complex geometric opinions.

Abstract

Collective decision making is often a customary action taken in government crowdsourcing. Through ensemble of opinions (popularly known as judgment analysis), governments can satisfy majority of the people who provided opinions. This has various real-world applications like urban planning or participatory budgeting that require setting up {\em facilities} based on the opinions of citizens. Recently, there is an emerging interest in performing judgment analysis on opinions that are constrained. We consider a new dimension of this problem that accommodate background constraints in the problem of judgment analysis, which ensures the collection of more responsible opinions. The background constraints refer to the restrictions (with respect to the existing infrastructure) to be taken care of while performing the consensus of opinions. In this paper, we address the said kind of problems with efficient unsupervised approaches of learning suitably modified to cater to the constraints of urban planning. We demonstrate the effectiveness of this approach in various scenarios where the opinions are taken for setting up ATM counters and sewage lines. Our main contributions encompass a novel approach of collecting data for smart city planning (in the presence of constraints), development of methods for opinion aggregation in various formats. As a whole, we present a new dimension of judgment analysis by adding background constraints to the problem.

Figures (1)

• Figure 1: An example scenario of judgment analysis with background constraints is shown. The outside boundary denotes the area under consideration for sewage planning and construction. The problem aims to set up new sewage lines in a city by taking suggestions (opinions) from the crowd. We consider that each crowd provides two opinions in this regard. Judgments are to be derived by computing consensus of the suggested sewage lines (shown as dashed lines) opined by the crowd. Background constraints are related to the sewage lines already in existence (shown as solid lines). In this example, the background constraint is: (1) Each suggested sewage line must have at least one intersection point with the existing sewage lines, and the additional general constraints are: (2) Each suggested sewage line must have a maximum length, and (3) A pair of suggested sewage lines must have a minimum distance between themselves. Note that the opinions given by the crowd 4, 5 and 6 satisfy all the constraints and are valid. However, the opinions of crowd 1 does not satisfy the constraints (2) and (3), the opinions of crowd 2 does not satisfy the constraints (1) and (2), and the opinions of crowd 3 does not satisfy the constraints (1) and (3). Hence, the final judgment is obtained from the consensus of valid opinions provided by the crowd 4, 5 and 6 only.