Interesting question, I was looking for something similar but found nothing. Now, we are working on an ontology incorporating the Wisdom-of-Crowds in some way and finishing the paper.I think there might be some similarities.
There are 3 dimensions in the question: the ontology manager system, the ontology encoding of the knowledge and the collective intelligence related embedded knowledge. which one are you focussing on ?
We have done some work on ontology encoding of the knowledge but in a different context. We analysed the context in which users need information and use that as part of the encoding scheme. Now we are trying to capture user action as result of getting the information, aggregate captured user actions from many users using STT data model and add this information back to the ontology. Thus broader aim is to enhance collective intelligence. But the work is still at an early stage.
The project that we are working on is to develop a mobile based information system for farmers (www.sln4mop.org). Farmers need agricultural information within their own context. We have identified important aspects of their context and found that can be modelled using following dimensions; Farm environment, Types of farmers, Preferences of farmer and Farming stages.
Related scenarios and use cases are described in the following papers;
Building Social Life Networks through Mobile Interfaces‒ the Case Study of Sri Lanka Farmers
User Centered Scenario based Approach for Developing Mobile Interfaces for Social Life Networks
Full text of these two papers are available on the Research Gate.
Yes, we used local languages for the field trial that we ran in December 2012. We used low cost (less than $100) Android Smart phones running "ice cream sandwich". This version provides UNICODE support. We had to install the local fonts. Ontology and other back end processing used English. We had a text file listing local language strings corresponding to English strings. Just before displaying the text on mobile interfaces we did the swap.
User input was limited to selecting predefined strings / symbols or entering numbers to specify cost, production quantity etc.
In our project we focus on a basic ontology which has a fixed structure but can be populated by the crowd. There are properties, which the crowd can choose from and update the ontology. As an example, we chose a tourism ontology for attractions. Tourists can add instances and vote for predefined properties, e.g. perceptions. In this way we get a shared view on attractions and how they are perceived by the public.
Your question has just slipped through, I wasn't ignoring you :). There's no final dataset in this project, because the system uses a dynamic ontology incorporating the user clicks on perceptions on the fly. Every moment the relative importance of a perception might be changed by someone. With this we can establish a dynamic Web site highlighting characteristics of a place according to the users' perceptions. The only validation, I admit, is performed by the localisation component (GPS), which checks whether the user is actually at the place or not.