Monday, 10 June 2013

Challenges for Volunteered Geographic Information: Credibility and Quality

by Benjamin Herfort

Quite a lot people say that we are living in an information age. Digital network technologies have reduced the costs of information production and dissemination in a stunning way so that it seems almost as if information is available nearly to everyone and everywhere. The changes VGI causes for geography as an academic discipline are fundamental.

To understand the changes more profoundly, it is necessary to discuss briefly what credibility means. Credibility is generally thought to be the believability of a source or message and is composed of the two dimensions trustworthiness and expertise. It is based on the combination of both dimensions and their objective and subjective characteristics. (Flanagin 2008)

There are approaches that focus on expertise and that understand quality as an objective property of the information. They regard credibility as accuracy while other approaches deny that. They describe credibility as perception. They state that the same piece of information may be judged differently by different people and that credibility depends on the subjective perception of the information receiver. While the first approach is adequate for describing and evaluating scientific knowledge production and bases primarily on facts, the second fits better for VGI concerning political or social issues, where opinions or perspectives are central. (Flanagin 2008)

Academic research on VGI and its credibility mainly deals with the quality of Open StreetMap data and focuses on credibility as accuracy. Haklay (2010) examines the positional accuracy of an object in a given area and the correlation with the number of contributors. The more contributors participate the better the data quality evolves. This is hardly surprising and applies to open source projects in general as well. Girres and Touya (2010) extend the elements of spatial data quality and also study attribute accuracy, completeness, logical consistency, semantic accuracy, temporal accuracy, lineage and usage. On the one hand they compare the OSM data with an external reference dataset; on the other they examine the OSM dataset intrinsic, e. g. on inter-theme consistency. The major problem for data quality they note is the lack of specifications. OSM contributors are merely advised to follow the specifications but do not have to. Furthermore they criticize the heterogeneity of the data and the management of updates, because it is not systematic and depends on the interest of the contributors.

Examining data of political or humanitarian crisis reference databases that are compiled through government agency, it becomes evident that it is either not available or simply useless, often because of manipulation by the government itself. In addition in many cases it is practically impossible to decide whether a piece of information is true or false. Meier therefore chooses a different approach. He regards crowdsourcing as a form of non-probability sampling. Because the information you get might not be representative for the whole population further verification is inevitable. The platform Ushahidi for example has established the following criteria to mark a report as verified: information comes from multiple reliable sources; two or more messages from different phone numbers reporting about the same incident; further messages from twitter, email, etc. and a text message confirming the report; photo or video documentation of the report; direct contact to the person reporting. (

Obviously there are more approaches verifying information and more problems concerning this process than I presented to you. But when you are discussion about specifications and the insufficient credibility and quality of Volunteered Geographic Information, you should keep in mind what Meier says:

False information can cost lives, but no information can also cost lives, especially in a crisis zone. Correct information that arrives too late is useless.


Flanagin, A.J. & Metzger, M.J., 2008. The credibility of volunteered geographic information. In S. Elwood, hrsg. GeoJournal. Dordrecht: Springer Netherlands, S. 137-148.

Girres, J.-F. & Touya, G., 2010. Quality Assessment of the French OpenStreetMap Dataset. In Transactions in GIS. Oxford: Blackwell Publishers Ltd, S. 435.

Haklay, M. (Muki) u. a., 2010. How Many Volunteers Does it Take to Map an Area Well? The Validity of Linus’ Law to Volunteered Geographic Information. In Cartographic Journal, The. S. 315-322.

Meier. Verifying Crowdsourced Social Media Reports for Live Crisis Mapping: An Introduction to Information Forensics. Online unter:


  1. I think the real challenge about credibility and quality is an automatic rating. The Ushahidi approach seems for me time-consuming, not the best condition for crisis response.

  2. That's a big issue, of course. The Ushadidi approach is quite similar to traditonal journalistic methods. Although it's efficient to me, when there are many contributors.

    There are some authors(e.g. Kumar et al.), that try to rate the quality of (Twitter)messages using a scoring system , that uses personal information about the contributor, geographic position and time.