From participatory sensing to participatory interpretation

The wide adoption of sensor-rich mobile phones in combination with the spread of web 2.0 culture recently created the right conditions for a new scope of research referred to as participatory sensing, which comes to complement previous efforts in sensor networks by involving directly the public and enabling new low-cost and large-scale practices of observation. Despite this new sensing capacity , a traditional issue remains: how to process sensor data into actionable knowledge? This asymmetry between this new sensing capacity and the lack of sense-making is not sustainable.. The volume and the numerical nature of data is not suited for easily making sense for the public, new producer and user of data.

One year ago, I started to investigate a new collective intelligence approach (human -social tagging- +machine -classifiers-) to generate and associate contextual & semantic information to sensor data, by integrating contextual tags and open annotation system, projecting raw data from a numerical space into a semantic one in order to prepare and support representation for different stakeholders and non-experts.

Despite great advances, many sensor network systems do not fully manage to process the amount of raw sensor data into actionable knowledge in order to be practically useful. Such problem is even more crucial in the case of participatory sensing where the public, producer and user of those data, has a low level of expertise to interpret them. The numerical and abondance nature of such sensor data is not well suited for easily making sense and thus inferences for humans (and machines).–>

Here is a quite old presentation (= doesn’t fully integrate all the concepts I have) about environmental tagging and the start of the above idea I had during the EU project Tagora on semiotics dynamics.For a demo , check also the semantic tool in the NoiseTube project, a visualization tool using a multi-dimensional “Tag cloud” representing any dataset in this semantic space by showing the co-occurence networks between contextual tags and thus facilitating the discover of contextual patterns.

Recently I decided to rework on this subject, still in progress, to consolidate and extend it.Currently the semantic space is static: The structure and contextual dimensions (time, space, social&user, noise, weather) are predefined. Instead of starting from the assumption that the goal of information systems is to process correct, context&free representations, a more fruitful approach would be to approach the challenge with a flexible mind and recognize that any resource will mean different things to different people in different contexts. The structure and there related tags can’t be static. With the spirit of constructivistic and pluralistic view, the notion that “it is we ourselves who create categories and force reality into supposedly insular compartments” (Zerubavel, 1991), the goal will be to design information systems that accommodate users from diverse societies, cultures, and understandings, the design of new open system empowering viewers to project massive amount of raw data into their own representation (e.g. personal semantic space, => the tag cloud representing the dataset should changes regarding the user profile).

I will try to publish a paper soon. (if your are interested in such research , contact me)