Ontologies for Quantified Self: a semantic approach Federica Cena, Silvia Likavec, Alessandro Marcengo Telecom Italia – Research and Prototyping Amon Rapp, Martina Deplano Department University of Turin Via Reis Romoli 274, Torino, Italy Corso Svizzera 185, Torino, Italy alessandro.marcengo@telecomitalia.it {cena,likavec,rapp,deplano}@di.unito.it ABSTRACT places that the users have been to, things they have seen, how they The spreading of devices and applications that allow people to sleep, how active they are, etc., creating a constant stream of data collect personal information opens new opportunities for user that can reveal many aspects of their lives. modeling (UM). In this new scenario UM together with personal However, today all these data are scattered in autonomous silos informatics (PI) can offer a new way for self-monitoring that can and not integrated. UM techniques have the potential of provide the users with a sophisticated mirror of their behavior, aggregating and correlating data not only coming from web attitudes and habits and their consequences on their life, on the browsing but also provided by all these PI systems. A UM environment and on contexts in which they live in. These new enriched with a plethora of personal data (behavioral, forms of self-reflection and self-knowledge can trigger and psychological, physical and environmental), related to different motivate the behavior change. In this paper we describe the first aspects of a person’s daily life, will be able to provide the user step in this direction, focusing on opportunities offered by with a “mirror” of herself, a sophisticated representation of semantic web ontologies for data integration and reasoning over interests, habits, activities in her life, in a novel way that is not yet data for recommendation purposes. achieved by any of the personal informatics tools available today [2]. This can support a new complex form of self-awareness and Categories and Subject Descriptors self-knowledge, which could foster behavior change processes [3], promoting more sustainable or healthier behavior, H.5.m. Information interfaces and presentation (e.g., HCI): discouraging bad habits, sustaining therapeutic improvement and Miscellaneous managing chronic diseases. In this new scenario UM together with PI can offer a new way for General Terms self-monitoring people’s own behavior, where self-monitoring Languages. refers to an assessment strategy to increase a person's awareness of targeted behavior [4], in order to promote behavior change [5]. Keywords UM and PI can provide users with a sophisticated mirror of their Ontologies, User model, Personal informatics, Quantified Self. behavior, attitudes and habits, highlighting their consequences on their life, on the environment and on contexts in which they live 1. INTRODUCTION in, promoting a new form of self-reflection and self-knowledge Personalized systems are used to meet individual preferences and that can trigger and motivate the behavior change. needs of each specific user, thus tailoring the system response to Our goal is to design a sophisticated UM-based PI system which these particular requirements. Personalized systems extrapolate can: users’ interests and preferences from explicit user ratings and from the observation of user behavior on the web: the system's i) gather heterogeneous types of user data (from PI systems' assumptions about the user based on these observations are stored sensors, from social web activities, from user’s browsing in a User Model (UM) [1]. A user model is the repository of behavior) and integrate them in an enhanced UM; personal information that has the potential to drive personalization ii) reason on the gathered data in order to find aggregations and and learning. The UM contains different types of information: correlations among data; from user demographic data to domain-specific preferences data iii) provide users with recommendations and meaningful UM (interest, knowledge…). visualizations to support self-awareness and self-knowledge. On the other hand, Personal Informatics (PI), also known as The paper is structured as follows. We first present our solutions Quantified Self (QS), is a school of thought which aims to use the and then we focus on semantic modeling of the domain in order increasingly popular invisible technology means for acquiring and to allow data integration and reasoning. collecting data on different aspects of the daily lives of people. They allow users to self-track a variety of data about their own behavior: these data can be, on the one hand, user physical states 2. STATE OF THE ART (such as glucose level in the blood), psychological states (such as Traditionally, User Models (UMs) [1,6] have the following mood), behavior (such as movements), habits (such as food features: (i) they are restricted to a single application; (ii) data are intake, sleep); on the other hand, they can be environmental derived from the web; (iii) they concern short periods of time. parameters (such as CO2 content, temperature) and contextual With the advent of ubiquitous computing technologies we are able information (such as people meeting) of the places passed through to track and store large amounts of various personal information, by the users during their everyday life. Thus, with this technology, scattered among applications and not integrated [7] even though it we have the capability to automatically record at large scale the is possible to integrate them with semantic web techniques [8]. In this paper we focus on data integration and reasoning over data This project will advance the UM state of the art in the following: (points i) and ii)) exploiting opportunities offered by semantic web ontologies [23]. Another challenging issue, namely gathering • the integration of data derived from everyday life, in user data, is out of scope of this paper addition to the data derived from the web; • reasoning on that data to gain further correlations about user behavior. 4. ONTOLOGIES FOR QUANTIFIED SELF In order to be able to: The opportunity is related to obtaining a Lifelong user model that stores user information for a long period of time and is able to integrate heterogeneous data coming from different devices and manage user interest change [9]. This project is a first step in this sources direction. reason on these data in order to provide meaningful visualization and recommendation According to [10], an ontology can be seen as a ‘‘formal, explicit specification of a shared conceptualization’’. With explicit we design and develop three ontologies, modeling the three main specifications of domain objects and their properties, as well as concepts of the Quantified Self world: time, place and user the relationships between them, ontologies serve as powerful activities. Vital parameters such as weight, blood pressure or formalisms for knowledge representation, providing exact blood sugar content are also important parameters, but we omit semantics for each statement and avoiding semantic ambiguities. them from the preset analysis, since they are used primarily by For these reasons, ontologies are often used for semantic data medical experts and are hard to analyze by ordinary people. integration and for resolving semantic conflicts, as in Time ontology. We want to model the time from a user point of [11,12,13,14,15]. Also, the associated rigorous mechanisms allow view, distinguishing work days, weekends and holidays (religious for different forms of reasoning (for example, to deduce implicit and civil ones), as well as dividing each day into meaningful slots classes), as in [16,17]. (morning, afternoon, evening, night). Measuring users' daily affective experiences is an important way Place ontology. Again we want to model the place from a user to quantify their life. In [18], the authors measure users' emotions perspective, labeling the places where the user lives, works or at various moments throughout the day. They asked the users to does the activities, dividing them into indoor (school, house, gym, answer demographic and general satisfaction questions, to work, cinema, restaurant, etc.) and outdoor (park, street..). (See construct a short diary of the previous day, and then to answer Figure 1: Place ontology.) structured questions about each episode. In [19], the authors investigate digital recordings of everyday activities, known as visual lifelogging, and elaborate the selection of target activities for semantic analysis. They investigate the selection of semantic concepts for life logging which includes reasoning on semantic networks using a density-based approach. Motivating behavior change towards a more active lifestyle is a psychological, social and technological challenge. Several Personal Informatics Systems have been developed in order to try to modify a behavior by means of self-monitoring, such as Figure 1: Place ontology [20,21,22] Activities ontology. We tried to model all the user activities, dividing them into two main categories: activities with place 3. A NOVEL SEMANTIC PI SYSTEM change (such as transportation or sports with place change) and We design a novel enhanced PI system, integrated in people’s activities with no place change (such as sports with no place everyday lives, able to gather data in a transparent way and to change, intellectual activity, physical work, resting activity or build and maintain a sophisticated user model able to aggregate feelings). Each of these classes has additional subclasses to better data and provide meaningful visualization and personalized describe the performed activity, but we omit them from the recommendations to the user for promoting behavior change. To picture for better clarity. For example, sports with place change reach this goal, we need the following components: has as its subclasses running, cycling, kayaking or downhill i) data integration of different user data for building a skiing, to name just a few. The design of this ontology was sophisticated model of user behavior, habits, needs and motivated by the categorization of activities in “Moves” preferences coming from different sources (web and real life application (https://www.moves-app.com). (See Figure 2: behavior) Activities ontology.) For lack of space, we included feelings into “Activities ontology”. We actually intend to have an additional ii) advanced forms of reasoning on user data for correlating different aspects of user daily behavior “Wellbeing and emotions ontology” to model user’s emotional state and wellbeing, taking inspiration from [24]. iii) personalized feedback for triggering behavior change in the users: • recommendations triggered by the correlation of different types of data (e.g., recommendations in accordance with user behavior, attitudes and habits in the UM) • meaningful visualization of data for raising awareness and motivating people in changing their behavior. for modeling the core aspects of user behavior which would help overcome these problems. This work is still at its early stage. We aim at experimentally evaluating our proposal by means of user tests to see short and long term effects of recommendations and visualizations on user behavior, as well as the acceptability of the solution. 6. ACKNOWLEDGMENTS Our gratitude goes to Telecom Italia for their support. Figure 2: Activities ontology 7. REFERENCES [1] Brusilovsky, P.: Methods and techniques of adaptive Then, we use these ontologies in two ways. hypermedia. User Modeling and User-Adappted Interaction, First, we use ontologies to solve the possible data value and v 6, n 2-3, pp 87-129, 1996 schema conflicts occurring among the data gathered from PI tools. [2] Li, I., Dey, A. K., Forlizzi, J.: A Stage-Based Model of As an example of data value conflicts, we gather “steps” both Personal Informatics Systems. In Proceedings of SIGCHI from the pedometer on the smart phone and from the smart Conf. on Human Factors in Computing Systems, pp. 557- bracelet and the collected numbers can differ: thus, in this case, 566. ACM, NY, USA, 2010 we calculate an average number of steps. Even more challenging would be to deal with contradicting or seemingly unrelated data. [3] Bandura, A.: Social Cognitive Theory of Self-Regulation. For example, a pedometer might suggest that you were sedentary, Organizational Behavior and Human Decision Processes, while at the same time having the gym as your location. 50, 248--287 1991 Pedometer forgotten in the locker or sitting in the gym bar? [4] Burke, L. E., Wang, J., Sevick, M. A.: Self-monitoring in Another example concerns the mood levels: from an ad hoc app weight loss: a systematic review of the literature. Journal of on the smart phone, we gather 4 mood values, whereas from the the American Dietetic Association, 111(1), 92-102, 2011 tangible channel we gather 6 mood values. Hence, the values [5] Bertram, C. L., Wang, J. B., Patterson, R. E., Newman, V. should be normalized. A., Parker, B. A., Pierce, J. P.: Web based self monitoring for Schema conflicts are more complex: for example, what is weight loss among overweight/obese women at increased modeled as an attribute in one relational schema may be modeled risk for breast cancer: the HELP pilot study. Psycho- as an entity in another schema (e.g. "hour" as an attribute for the Oncology, 22(8), 1821-1828, 2013 entity "sleep" and "hour" as an entity that has a relationship with [6] Heckmann, D., Schwartz, T., Brandherm, B., Schmitz, M., "sleep"). As another example, two sources may use different von Wilamowitz-Moellendorff, M.: Gumo - the general user names to represent the same concept (e.g. "running" and model ontology. In User Modeling 2005, pp. 428-432, "jogging"), or the same name to represent different concepts, or Springer, 2005 two different ways for conveying the same information (e.g. "date of birth" and "age"). We solve these conflicts by mapping the data [7] Aroyo, L., Dolog, P., Houben, G.-J., Kravcik, M., Naeve, A., to our ontologies. Nilsson, M., Wild, F.: Interoperability in personalized adaptive learning. Journal of Educational Technology and Second, we use these ontologies to make inferences useful for Society 9(2), 4–18, 2006 recommendation, in conjunction with Data Mining techniques for discovering correlations among data, where various forms of [8] Sluijs, K.van der, Houben, G.-J.: A generic component for generalization can make correlations more powerful. For example, exchanging user models between web-based systems. data mining techniques might provide a correlation between International Journal of Continuing Engineering Education headache and running or biking activities. Since the two activities and Life-Long Learning 16(1–2), 64–76, 2006 are two types of "outdoor activities" in the Activities Ontology, [9] Kay, J., Kummerfeld, B. eds. Proceedings of the Lifelong we can indicate a correlation between outdoor activities and User Modelling Workshop, at User Modeling Adaptation and headache. Alternatively, if we know that a certain user has a Personalization Conf. UMAP '09, 2009 headache on December 24th, January 1st and August 15th, and from the Time Ontology we know that these are holidays, we can [10] Gruber, T. R.: A translation approach to portable ontology infer a correlation between holidays and headache. specifications. Knowledge Acquisition Journal 5 (2), 199– 220, 1993. Moreover, we could suggest a behavior that is similar or different but somehow related to what the user is used to doing. For [11] Arens, Y., Ciiee, Y., Knoblock. A.: SIMS Integrating data example, if we know that the user loves running, but according to from multiple information sources. Information science institute, University of Southern California, U.S.A, 1992 our data, we discovered that this is correlated with bad sleep, we might suggest some similar activities (in the same category) such [12] Goh, C.H., Bressan, S., Madnick, S. and Siegel. M.: Context as hiking or walking.. interchange New features and formalisms for the intelligent integration of information. ACM Transaction on Information Systems, 17(3):270–290, 1999 5. CONCLUSIONS In this paper we tackle an important problem of long term [13] Beneventano, D., Bergamaschi, S., Guerra, F. Vincini. M. management of users' data in PI systems and address a number of 2001: The MOMIS approach to information integration. In challenges including the need for data integration and ICEIS 2001, Proceedings of the 3rd Int. Conf. on Enterprise interpretation. We motivate the introduction of suitable ontologies Information Systems. [14] Visser, P. R., Jones, D. M., Beer, M. , Bench-Capon, T., [20] Shumaker, A., Ockene, J. K., Riekert, K.: The Handbook of Diaz, B. and Shave, M.: Resolving ontological heterogeneity Health Behavior Change, Springer, 2008 in the KRAFT project. In 10th Int. Conf. and Workshop on [21] Froehlich, J., Dillahunt, T., Klasnja, P., Mankoff, J., Database and Expert Systems Applications DEXA’99, 1999 Consolvo, S., Harrison, B., Landay, J.: UbiGreen: [15] Abel, F., Herder, E., Houben, G. J., Henze, N., Krause, D.: Investigating a Mobile Tool for Tracking and Supporting Cross-system user modeling and personalization on the Green Transportation Habits. In Proceedings of the SIGCHI social web. User Modeling and User-Adapted Interaction, Conf. on Human Factors in Computing Systems. pp. 1043- 23(2-3), pp.169-209, 2013 1052. ACM, New York, USA, 2009 [16] Wang, X. H., Zhang D. Q., Gu, T., Pung, H., K.: Ontology [22] Kay, M., Choe, E. K., Shepherd, J., Greenstein, B., Watson, Based Context Modeling and Reasoning using OWL. In N., Consolvo, S., And Kientz, J. A.: Lullaby: a capture & Proceedings of the 2nd IEEE Ann. Conf. on Pervasive access system for understanding the sleep environment. In Computing and Communications Workshops (PERCOMW Proceedings of 2012 ACM Conf. on Ubiquitous Computing. '04). IEEE Computer Society, 2004 pp. 226--234, ACM, New York, USA, 2012 [17] Eiter, T., Ianni, G., Polleres, A., Schindlauer, R., Tompits, [23] Guarino, N.: Formal ontology and information systems. In H.: Reasoning with rules and ontologies (2006), Reasoning Proceedings of the 1st Int. Conf. on Formal Ontology in Web 2006 Information Systems, FOIS ’98, IOS Press, pp 3-15, 1998. [18] Kahneman, D., Krueger, A.B., Schkade, D.A., Schwarz, N., [24] Patti, V., Bertola, F.: Organizing Artworks in an Ontology- Stone, A.A.: A survey method for characterizing daily life based Semantic Affective Space, Proceedings of the 1st Int. experience: The day reconstruction method, Science 306, pp. Workshop on Emotion and Sentiment in Social and 1776–1780, 2004 Expressive Media: approaches and perspectives from AI, [19] Wang, P., Smeaton, A.F.: Semantics-based selection of ESSEM@AI*IA, 2013. everyday concepts in visual lifelogging, International Journal of Multimedia Information Retrieval 1, pp 87–101, 2012