《Crowdsourcing Maps for Emergency Planning》由会员分享，可在线阅读，更多相关《Crowdsourcing Maps for Emergency Planning（10页珍藏版）》请在金锄头文库上搜索。
1、Collabmap: Crowdsourcing Maps for Emergency PlanningSarvapali D. Ramchurn,1 Trung Dong Huynh,1 Matteo Venanzi,1 Bing Shi21Electronics and Computer Science, University of Southampton, United Kingdomsdr,tdh,mv1g10ecs.soton.ac.uk2School of Computer Science and Technology, Wuhan University of Technology
2、Wuhan, CABSTRACTIn this paper, we present a software tool to help emergencyplanners at Hampshire County Council in the UK to createmaps for high- delity crowd simulations that require evac-uation routes from buildings to roads. The main featureof the system is a crowdsourcing mechanism that breaksdo
3、wn the problem of creating evacuation routes into micro-tasks that a contributor to the platform can execute in lessthan a minute. As part of the mechanism we developeda concensus-based trust mechanism that lters out incor-rect contributions and ensures that the individual tasks arecomplete and corr
4、ect. To drive people to contribute to theplatform, we experimented with di erent incentive mecha-nisms and applied these over di erent time scales, the aimbeing to evaluate what incentives work with di erent typesof crowds, including anonymous contributors from AmazonMechanical Turk. The results of
5、the in the wild deploymentof the system show that the system is e ective at engagingcontributors to perform tasks correctly and that users re-spond to incentives in di erent ways. More speci cally, weshow that purely social motives are not good enough to at-tract a large number of contributors and t
6、hat contributorsare averse to the uncertainty in winning rewards. Whentaken altogether, our results suggest that a combination ofincentives may be the best approach to harnessing the maxi-mum number of resources to get socially valuable tasks (suchfor planning applications) performed on a large scal
7、e.Categories and Subject DescriptorsC.5 World Wide Web: Crowdsourcing1. INTRODUCTIONThe creation of high delity scenarios for disaster simulationis a major challenge for a number of reasons. First, in theUK, the maps supplied by existing map providers (e.g., Ord-nance Survey, TeleAtlas) tend to prov
8、ide only road or build-ing shapes and do not accurately model open spaces whichpeople use to evacuate buildings, homes, or industrial facili-ties (e.g. the space around a stadium or a commercial centreboth constitute evacuation routes of di erent shapes andsizes). Secondly, even if some of the data
9、about evacuationroutes is available, the real-world connection points betweenthese spaces and roads and buildings is usually not well de-ned unless data from buildings owners can be obtained(e.g. building entrances, borders, and fences). Finally, inorder to augment current maps with accurate spatial
10、 data,it would require either a good set of training data (which isnot available to our knowledge) for a computer vision algo-rithm to de ne evacuation routes using pictures (working onaerial maps) or a signi cant amount of manpower to directlysurvey a vast area.Against this background, we developed
11、 a novel model ofgeospatial data creation, called CollabMap1, that relies onhuman computation. CollabMap is a crowdsourcing tool toget users to perform micro-tasks that involve augmenting ex-isting maps (e.g. Google Maps or Ordnance Survey) by draw-ing evacuation routes, using satellite imagery from
12、 GoogleMaps and panoramic views from Google StreetView. In asimilar vein to 12, 4, we use human computation to com-plete tasks that are hard for a computer vision algorithm toperform or to generate training data that could be used bya computer vision algorithm to automatically de ne evac-uation rout
13、es. Compared to other community-driven plat-forms such as OpenStreetMap and Googles MapMaker, Col-labmap allows inexperienced and anonymous users to per-form tasks without them needing the expertise to integratethe data into the system (as in OpenStreetMap) and doesnot rely on having experts verifyi
14、ng the tasks (as in Map-Maker) in order to generate meaningful results.To ensure that individual contributions are correct and com-plete, we build upon the Find-Fix-Verify (FFV) pattern 1to develop a novel adaptive work ow that includes concensus-based trust metrics and allows the creation of new ta
15、skswhere no ground-truth is known. Our trust metrics allowusers to rate and correct each others contributions while ourwork ow is adaptive in the sense that it allows the systemdesigner to improve the performance of the crowd accord-ing to both the number and types of contributions into thesystem. A
16、s we show in our results, this approach was ef-fective in preventing workers from getting bored and takingfull advantage of users motivation to contribute.Given our implementation of the platform, we deployed our1www.collabmap.orgsystem to help map the area around the Fawley Oil re n-ery next to the city of Southampton in the UK over threemonths. The area covered over 5,000 buildings (mainly res-idential) with a population of about 10,000. We experi-m
