Information dissemination on the Web starts with people interacting with existing applications. Hence, the needs of these people and they way they use these applications must be considered in any effort to improve the Web. This is done by characterizing and modeling user behavior patterns, quantifying and qualifying their interaction with applications. Such characterization and modeling go beyond the interaction between user and system, and consider users’ interactions with each other, thus creating social networks. Research in user behavior and interaction modeling (UBIM) develops techniques to provide input to the design of mechanisms that contribute to improving applications offered to users.
Research on qualitative aspects involves determining which applications people will want to use, will be able to use, and will consider effective when using . It also involves investigating social aspects of application usage, regarding the impact these applications have on people and on their relationships [1, 13, 98, 107]. As an example of such an impact, experiments could verify if people behave ethically when using a given application .
User behavior and their interaction can also be analyzed from a quantitative point of view. The quantitative characterization process may be based on various sources of evidence collected from the applications [5, 9, 38, 64, 114] and from observed traffic in distinct points of the network [61, 92]. Examples of such evidence include information available to users and accessed by them, data on interaction with other users, frequency of use of various applications, and visitation patterns. It is necessary to characterize such evidence, subsequently summarizing it into models that describe patterns of representative behavior. For that purpose, researchers may use data mining techniques [97, 112], statistical modeling , and Markovian and non-Markovian modeling , to accurately capture properties of static and dynamic behavior. Such techniques may also be employed in tandem with social network modeling strategies (SN research line), aiming at developing collective behavior models and discovering clusters (or communities) that share common interests. Information retrieval techniques (IR research line) may also be used to identify and classify behavior profiles [17, 64, 83].
In the context of our research program, user behavior and interaction modeling is strongly related to Challenges 1 and 3. All the goals associated to Challenge 1 are the focus of the investigation for this research line, in close collaboration with the social network research line (SN). The identification, characterization, and modeling of malicious, opportunistic and anti-social behavior patterns (goal 1.3) aim at creating the basis for the design of mechanisms to deal with this type of behavior (Challenge 3). Quantitative modeling of these patterns and the design of control mechanisms can benefit from the investigation of qualitative issues, such as impact on user actions and behavior (goal 1.5). Also, investigating how social behavior influences and is influenced by the dynamic characteristics of applications (goal 1.4) can be relevant to quantitative modeling.
User behavior and interaction modeling is also directly related to Challenge 3, since its focus includes delivery of information to users, quality of interaction, and social aspects that allow the delivery to be viewed as satisfactory. Therefore, goals 3.6 and 3.8 are directly related to this research line, especially regarding information visualization [33, 105], design and assessment of collaborative systems  and software development by end users [3, 56, 57].
The effort to associate qualitative and quantitative aspects in modeling is new, since these perspectives are usually considered independently. Such a combined perspective aims at capturing the main aspects regarding real user behavior in Web applications and producing models that are more representative and realistic. These models will support the investigation of the impact of applications on user behavior, as well as on the impact of behavior patterns on applications and infrastructure.
The User Behavior and Interaction Modeling (UBIM) research line will be carried out, primarily, by researchers Jussara Almeida (UFMG), Virgílio Almeida (UFMG) and Raquel Prates (UFMG).