Enabling efficient, supportive and pleasant user-adapted communication
The User-Adapted Communication and Ambient Intelligence Lab is focused on innovating user interaction to context-dependent communication applications and services. Across a wide span of applications (including telecommunications, multimedia access, learning, and an intelligent user environment), our central goal is to improve the service brought to the end-user in terms of her/his experience, satisfaction, efficiency and gain. Our interest is to bring a solution to the end-users that is adapted to real-world challenges.
The following challenges are to be addressed on the way to the user-adapted solutions:
Human-machine communication is context dependent. End user requirements, the ways he/she makes decisions and using services are dependent on several contexts, such as place, social interaction etc. Any effective solution should utilize any available contextual information.
The natural user-to-machine communication is dependent on user’s personality, mood and emotions. The way how he/she makes decisions and his/her expectations of communication-device responses depends on his/her personality, his/her current emotions and mood. The communication solution should thus incorporate these individual and user’s state differences.
The user-to-machine communication is continuous. The natural way of communication with devices is the one used in human-to-human communication. Among other aspects, this communication is continuous over time, with an immediate reaction to most of the information provided by the other party. Therefore, the communication system should be conversational in terms of the next step being fully aware of all the previous steps of the current session. These steps must progress in a natural, expected way for each individual user.
The communication system is autonomous in information gathering and intelligent decision. The communication system should gather information about the user and about the context in an unobtrusive way and process it autonomously to support the users. Besides, the active adaptation to each individual end user and to each specific situation required intelligence in terms of understanding complex data and in terms of predicting immediate user needs and decisions.
To make a step forward towards a solution of the above-listed challenges, our group is concentrating on the following fields.
User modelling and context-aware conversational recommender systems. The intelligence required to actively support the end-user in a real-time interfacing to devices lies in user-modelling-related machine-learning algorithms, including conversational recommender systems. An accurate prediction of the next step the end-user is most likely to take while interacting with devices is usually required in order to adapt this process to the user.
User interface design. As well known, the form of the user interface plays a crucial role in the process of interfacing with end-user devices. The optimal design is related to several other aspects of the whole interfacing process and as such is an interdisciplinary challenge.
Multimedia signals and communication. The content brought to end-users by modern communication services is in the form of multimedia. Thus, the rules and dynamics of multimedia communication should also be taken into account. We exploit video analysis for the detection of end-user’s emotional state, etc.
Social signals.Social signals are the most natural way of information exchange on his/her state of mind during the humans-to-human communication in real time. As such it has a great potential in non-intrusive user data acquisition in real time. The automatic recognition of social signals is based on multimedia signals analysis such as sound and video analysis.
Internet of things (IoT) and big/data analysis for end-user context identification.Several types of data required to identify the context of the user interfacing can be gathered automatically without any user attention. The rapid development of big-data analysis techniques provided even more effective methods for context identification that cannot be neglected. We use modern big-data analysis techniques to automatically identify the relevant contexts of end-users in order to further adapt the service.
Special communication hardware solutions. One aspect of user-adapted communication is that the solutions vary greatly among the different target user groups, among different types of applications, etc. Consequently, special-purpose hardware solutions as a part of the adapted user interfaces are required.
Social network user data acquisition. The data derived from end-user social-network-related behaviour is difficult or impossible to gather in any other way. Since the end-user’s record of behaviour spans a longer period of time, relevant behaviour patterns can be identified and utilized in user-adapted communication services.