Combining Symbolic and Data-Driven Methods for Goal Recognition
| Autor: | Christian Bartelt, Heiner Stuckenschmidt, Nils Wilken |
|---|---|
| Rok vydání: | 2021 |
| Předmět: |
Focus (computing)
Ubiquitous computing Computer science Reliability (computer networking) 010401 analytical chemistry 020206 networking & telecommunications Context (language use) 02 engineering and technology 01 natural sciences 0104 chemical sciences Data-driven Human–computer interaction 0202 electrical engineering electronic engineering information engineering Feature (machine learning) Task analysis Goal recognition |
| Zdroj: | PerCom Workshops |
| DOI: | 10.1109/percomworkshops51409.2021.9431025 |
| Popis: | Recently, there is an increased research interest in context-aware systems that are able to autonomously and intelligently support users with their tasks. An important feature of such systems is the ability to dynamically adapt to the current needs of their users. Hence, it is critical for such systems to be able to recognize the user's current activities and goals. While there is some work on the problem of goal recognition in this context, the majority of research works focus on the problem of recognizing a user's current activities. Furthermore, the existing methods mostly rely on purely symbolic methods, which have problems to handle low signals in the observed user data. As a consequence, these approaches are not able to reliably recognize the user goals as early as it should be possible, based on the information contained in the observed data. Hence, in our research, we focus on the combination of symbolic and data-driven methods to hybrid methods for goal recognition and their application in the context of pervasive computing environments like smart homes. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|