COMPASS: Unsupervised and online clustering of complex human activities from smartphone sensors

Campana, Mattia Giovanni; Delmastro, Franca

doi:10.1016/j.eswa.2021.115124

Cited by 4 publications

(4 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Machine learning techniques were adopted to discover hidden contexts from raw data (Livne et al, 2022;Unger et al, 2016). For example, Campana and Delmastro (2021) apply clustering algorithms to extract environmental context (e.g., location, nearby devices) from the mobiles. Afzal et al (2018), Chahuara et al (2017), and E. J. emphasize the separation between a low-level and high-level context for accurate reasoning and the capacity of reusability.…”

Section: Behavioral Aspectmentioning

confidence: 99%

Introduction to the Minitrack on Human-Computer Interaction in the Digital Economy

Schneider,

Valacich,

Jenkins

2024

Proceedings of the Annual Hawaii International Conference on System Sciences

View full text Add to dashboard Cite

Context awareness systems, a subcategory of intelligent systems, are concerned with suggesting relevant products/services to users' situations as smart services. One key element for improving smart services' quality is to organize and manipulate contextual data in an appropriate manner to facilitate knowledge generation from these data. In this light, a knowledge-based approach, can be used as a key component in context-aware systems. Context awareness and knowledge-based systems, in fact, have been gaining prominence in their respective domains for decades. However, few studies have focused on how to reconcile the two fields to maximize the benefits of each field. For this reason, the objective of this paper is to present a literature review of how contextaware systems, with a focus on the knowledge-based approach, have recently been conceptualized to promote further research in this area. In the end, the implications and current challenges of the study will be discussed.

show abstract

Section: Behavioral Aspectmentioning

confidence: 99%

Introduction to the Minitrack on Human-Computer Interaction in the Digital Economy

Schneider,

Valacich,

Jenkins

2024

Proceedings of the Annual Hawaii International Conference on System Sciences

View full text Add to dashboard Cite

show abstract

“…However, traditional (i.e., offline) clustering requires to keep all the data samples in memory and to process them several times to find the best clusters' configuration that optimizes a given metric (e.g., the average squared Euclidean distance in K-means), which is not feasible for memory-constrained devices like smartphones and wearables. Therefore, to preprocess the GPS data on the local device, the GPS based module relies on COMPASS, an online clustering algorithm we recently proposed in the literature to identify hidden patterns from mobile sensors data streams [24]. The algorithm has been compared with the state-of-the-art, both for offline and online clustering approaches, demonstrating to outperform the reference algorithms.…”

Section: Proximity Based Gps Basedmentioning

confidence: 99%

“…On the other hand, the execution time required to calculate ω a in the Familiar Places module mainly depends on the type of context data to process. While for proximity data Equation 4 has a constant time complexity O(1) because it simply incrementally updates the total time an alter has been in proximity of the local device, GPS data processing requires a pre-clustering phase, which can be efficiently performed in just a few milliseconds with an online clustering algorithm like COMPASS [24].…”

Section: Complexity Evaluationmentioning

confidence: 99%

“…In this case social context and location information are based on data derived from smartphone-embedded sensors and from the user interaction with her personal mobile device. As shown in Figure 1, we envision the implementation of the proposed models as part of a pre-existing middleware architecture [13], which includes all the necessary components to perform the context-recognition task directly on the user's device: a Sensing Manager (SM) layer that unobtrusively collects context data from both physical and virtual sensors available on mobile devices, a Context Modeling (CM) layer aimed at processing raw sensor data to extract meaningful features to characterize the user's context; and a Context Reasoning (CR) layer that relies on such features to recognize the user's context by using machine learning models, including both unsupervised clustering solutions [24,25] and pre-trained supervised classifiers (e.g., Random Forest and Artificial Neural Networks [26,27]).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

On-device modeling of user's social context and familiar places from smartphone-embedded sensor data

Campana¹,

Delmastro²

2022

Preprint

Self Cite

View full text Add to dashboard Cite

Context modeling and recognition represent complex tasks that allow mobile and ubiquitous computing applications to adapt to the user's situation. The real advantage of context-awareness in mobile environments mainly relies on the prompt system's and applications' reaction to context changes. Current solutions mainly focus on limited context information generally processed on centralized architectures, potentially exposing users' personal data to privacy leakage, and missing personalization features. For these reasons on-device context modeling and recognition represent the current research trend in this area. Among the different information characterizing the user's context in mobile environments, social interactions and visited locations remarkably contribute to the characterization of daily life scenarios. In this paper we propose a novel, unsupervised and lightweight approach to model the user's social context and her locations based on ego networks directly on the user mobile device. Relying on this model, the system is able to extract high-level and semantic-rich context features from smartphone-embedded sensors data. Specifically, for the social context it exploits data related to both physical and cyber social interactions among users and their devices. As far as location context is concerned, we assume that it is more relevant to model the familiarity degree of a specific location for the user's context than the raw location data, both in terms of GPS coordinates and proximity devices. We demonstrate the effectiveness of the proposed approach with 3 different sets of experiments by using 5 real-world datasets collected from a total of 956 personal mobile devices. Specifically, we assess the structure of the social and location ego networks, we provide a semantic evaluation of the proposed models and a complexity evaluation in terms of mobile computing performance. Finally, we demonstrate the relevance of the extracted features by showing the performance of 3 different machine learning algorithms to recognize daily-life situations, obtaining an improvement of 3% of AUROC, 9% of Precision, and 5% in terms of Recall with respect to use only features related to physical context.

show abstract

On-device modeling of user’s social context and familiar places from smartphone-embedded sensor data

Campana

Delmastro

2022

Journal of Network and Computer Applications

View full text Add to dashboard Cite

COMPASS: Unsupervised and online clustering of complex human activities from smartphone sensors

Cited by 4 publications

References 44 publications

Introduction to the Minitrack on Human-Computer Interaction in the Digital Economy

Introduction to the Minitrack on Human-Computer Interaction in the Digital Economy

On-device modeling of user's social context and familiar places from smartphone-embedded sensor data

On-device modeling of user’s social context and familiar places from smartphone-embedded sensor data

Contact Info

Product

Resources

About