|Title:||Social-Sensors Dataset: "NUS-SENSE"||Creators:||Farseev, A.
Chua Tat Seng
|NUS Contact:||CHUA TAT SENG||External Contact:||Subject:||Artificial intelligence
Social networking (online)
Body mass index
Individual user profile
Social media datum
Social media platforms
Wellness is a widely popular concept that is commonly applied to fitness and self-help products or services. Inference of wellness-related attributes (wellness user profiling), such as body mass index (BMI) or diseases tendency, is of crucial importance to various applications in personal and public wellness domains. At the same time, the emergence of social media platforms and wearable sensors makes it feasible to learn wellness user profile from multiple perspectives. However, the research efforts on wellness user profiling from multiple social sources are relatively sparse, while the joint analysis of social media and sensor data was not comprehensively studied yet. We introduce a large-scale dataset towards joint sensor and social media data analysis.
In order to build a comprehensive wellness profile, we harvested data of different modality from multiple social networks: Twitter micro-posts were used as a textual data source; Instagram pictures and it's descriptions (comments) were used as an image and textual data sources, respectively; Foursquare check-in records were used as a location data source; User's workouts were used as a sensor data source and for ground truth construction. The sensors data bridges the gap between social media-based users' representation and their actual physical condition.
Our dataset can be used for both descriptive and prescriptive research. That is to say, we do not intend to constraint future research on user profile learning, since the available ground truth provides possibility to tackle other contemporary problems. The potential research topics that can be conducted on our released dataset are listed below: Extended multi-source user profile learning. It could be useful to perform further modeling of multi-source multi-modal data. From the data point of view, it is interesting to gain deeper inside into visual and sensor data representations extraction. From the data modeling point of view, it is interesting to study the performance of advanced non-linear models on wellness profile learning.
For more details of this dataset and to reuse this dataset, please visit http://nussense.azurewebsites.net/.
|Citation:||When using this data, please cite the original publication and also the dataset.
|Appears in Collections:||Staff Dataset|
Show full item record
Files in This Item:
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.