Self-adaptive applications: on the development of personalized web-tasking systems

Abstract

Personalized Web-Tasking (PWT) proposes the automation of user-centric and repetitive web interactions to assist users in the fulfilment of personal goals using internet systems. In PWT, both personal goals and internet systems are affected by unpredictable changes in user preferences, situations, system infrastructures and environments. Therefore, self-adaptation enhanced with dynamic context monitoring is required to guarantee the effectiveness of PWT systems that, despite context uncertainty, must guarantee the accomplishment of personal goals and deliver pleasant user experiences. This paper describes our approach to the development of PWT systems, which relies on self-adaptation and its enabling technologies. In particular, it presents our runtime modelling approach that is comprised of our PWT Ontology and Goal-oriented Context-sensitive web-tasking (GCT) models, and the way we exploit previous SEAMS contributions developed in our research group, the DYNAM-ICO reference model and the SmarterContext Monitoring Infrastructure and Reasoning Engine. The main goal of this paper is to demonstrate how the most crucial challenges in the engineering of PWT systems can be addressed by implementing them as self-adaptive software.

Web Augmentation as a Technique to Diminish User Interactions in Repetitive Tasks

“…It also takes advantage of advanced natural language understanding algorithms to analyze the user utterance to trigger the VASTA automation scripts, as well as to determine the automation parameters for generalization. PWT [69] provides end-users with a way to customize their own searches in their favourite or most visited web pages, thus reducing frustration in repetitive tasks. Despite the existence of applications to automate and customize web interactions, these are insufficient to ensure that user goals are met when changes in the relevant context cannot be fully anticipated at design time [69].…”

Personalized Web-Tasking Applications: An Online Grocery Shopping Prototype

“…Recently we proposed two runtime modelling approaches [11]: (1) our Personalized Web-Tasking (PWT) model, and (2) our Goal-Oriented Context-Sensitive Web-Tasking (GCT) model. These runtime models define the elements and concepts of personalized web-tasking, as well as the representation of the personal goals of the user and relevant context.…”

“…Our runtime models (PWT model and GCT model) communicate with all three DYNAMICO layers to provide runtime adaptation capabilities (i.e., changes in the web-tasking conceptual elements, the PWT system infrastructure, and the system requirements). Finally, in the lower layer of our design we take advantage of the existing SMARTERCONTEXT Monitoring Infrastructure proposed by Villegas [12] to support contextawareness capabilities to our system [11]. The Web-Tasking Knowledge Infrastructure is the platform to understand the personal user goals.…”

“…To populate the information of the web-tasking knowledge infrastructure we defined two personal goals for the grocery shopping expressed in observable results as represented in our PWT model [11]: (1) The numbers of items left in the grocery shopping list application at the end of the webtasking execution should be less than 20% of the original.…”

“…We proposed a PWT application as a self-adaptive software system, based on the DYNAMICO reference model proposed by Villegas et al [10], and on our runtime models [11]: (1) our PWT model to represent the concepts of personalized web tasking, and (2) our goal-oriented contextsensitive web-tasking (GCT) model to represent the evolving web-tasking personal goals of the user, as well as the relevant context. Our PWT System comprises four software modules: (i) Web-Tasking Knowledge Infrastructure, which provides a mechanism to express personal user goals in the form of a GCT model; (ii) PWT Model Processor, to analyse our PWT models and generate the corresponding RDF graphs that contain the information about the user's webtasking; (iii) Personalization Engine, which uses personal context information to improve the execution of the webtasking; and (iv) Web-Tasking Effector, which executes the sequence of web interactions on behalf of the user while managing the life-cycle of the web-tasking .…”