Utskrift från Malmö universitets webbplats www.mah.se

Industrial Graduate School in The Data Driven systems, DDS

Contact person: Bengt J. Nilsson
Responsible: Bengt J. Nilsson
Co-workers: Andreas Jacobsson, Paul Davidsson, Åse Jevinger and Anders Kärrman
Partner: Apptus Technologies AB (Lund), Massive Entertainment (Malmö), SIGMA Technology Solutions AB (Malmö) och Sony Mobile (Lund)
Funding: KK-stiftelsen
Timeframe: 2018-03-01 -- 2021-12-31
Faculty/Department: Faculty of Technology and Society, Department of Computer Science and Media Technology

Data-driven systems

Industrial graduate school will address a research field that we have labelled “data-driven systems”. Automated and semi-automated processing of digital information is of growing concern and interest for society and industry alike. The emerging IoT-paradigm facilitates real-

world data collection while advances in Artificial Intelligence (AI) promises to automate or support decision making on the basis of such data. For such systems to be successful however, in many cases they need to offer a good balance between automation and human

intervention. Regardless of the specific application field in each particular case, the licentiate projects will typically address the challenge of automated and semi automated data-driven

systems from different perspectives, implying a necessary acquaintance with several areas of knowledge, including:


Data mining, which concerns the extraction of relevant information from large data sets in forms that are suitable for further use by humans and/or machines.

Within the area of Machine Learning, a large number of methods to automate this process have

been developed. Even though substantial progress has been made recently making the method more powerful, e.g. “deep learning”, it is not yet fully understood which method is the best one to use for a particular problem, as well as how to tune the parameters of the chosen algorithm.

Within DDS we will mainly focus on applying or adapting existing methods to novel problems.

An important application area for datamining within DDS is recommender systems, which

are automated tools and techniques that provide suggestions to a user. The suggestions relate to various decision making processes, such as what items to buy, what music to listen to, or what books to read. Another important application area within DDS is learning predictive models,

e.g. for predictive maintenance of machines or optimizing computer games.

Using data mining methods to support crowd-sourcing of documentation is also something we plan to investigate within DDS.


Context-aware systems adapt their behaviour based on changes in the environment to

better suit the situation they are acting in. Within context-awareness, a broad range of

techniques, methods, models, applications and middleware solution are recognized.

Still, context-awareness would benefit from expanding beyond the largely sensorbased and real-time focus that is dominating the field. For instance, through pattern analysis of historical data related to the context, prediction of future contexts is a promising new direction. Within DDS, the emphasis will be two-fold: (1) effective processing and use of real-time context data, and (2) applying data mining and machine learning techniques on historical data to identify context cues that may effectively inform real-time system behaviour and prediction of future user needs. Im-

portant applications areas of context-awareness within DDS includes personalization of recommendations and automation, as well as improving systems intelligence.


User experience (UX) design originates from the desire to shift focus from product

design driven by functional testing to that which is experienced during use-time. Es-

tablished evaluation methods was identified early as a challenge which has since led

to considerable emphasis being placed into development of methods to use, metrics of

relevance, and how to systematically work with UX design as part of product development. The impact of changing user needs depending on use situation, as well as the perception of themselves and their previous experiences changing over time, has recently been emphasized as a challenge. UX design has yet to identify suitable design patterns to negotiate this. In hopes of better understanding the user base, post-deployment data collection is becoming a mainstream strategy, despite the struggle of making sense of these rapidly growing datasets. Here, UX design has an opportunity to identify methods for continuous user experience evaluation of products as well as more dynamic metrics. Within DDS, the emphasis will be on identifying design para-

digms that effectively adapt products and services to the changing user needs over

time, and bridging these with product innovation and development strategies that em-

brace data collection and use pattern analysis as core competitive elements.


Many of the involved researchers in the environment have long experience from researching

these fields and have published their results in international peer-reviewed journals and con-

ferences (see CVs in appendix). An effect of DDS is that researchers with deep knowledge of different areas will come together, which will constitute an important asset for both the stu-

dents and their companies. Nationally, the researchers are deeply involved (board members or similar) in a number of relevant research organisations, such as, Swedish AI Society, Swedish Strategic Innovation Program for the Internet of Things, SwedSoft, Swedish IT Security Network for PhD Students, and National Research School of Intelligent Transport Systems.

We also have close collaboration with some Swedish universities that have strong research in the area of DDS, such as Blekinge Institute of Technology, Chalmers and the Royal Institute of Technology. Internationally, the researchers are deeply involved (board members or similar) in a number of relevant research organisations, such as, Europe an Research Centre of Network Intelligence for Innovation Enhancement, Multi-Agent-Based Simulation (The International Work-

shop Series), Practical Applications of Agents and Multi-Agent Systems (The International Conference Series), ARCO (Algorithmic Research Cooperation around Oresund) and NordiCHI (Nordic forum for Human-Computer Interaction research). Moreover, the researchers have been involved in hundreds of program committees of international scientific conferences relevant for DDS. Similarly, they continuously perform review work for many scientific journals relevant for DDS, and have performed (co-) editorial services for these journals.


Research question examples

Given the three research areas described above, there are a large number of research questions

that are relevant from both research and industry perspective. Most of them are not specific

for a particular application area, but are relevant for many business areas and thus provide good conditions for cross-company collaboration. Examples of research questions that are planned to be addressed in DDS graduate school include:


• How may data-driven approaches based on data from many different sources be used

and integrated in to product development and innovation?

• How may context-awareness be used as an approach to automation and improved user


• How may data from user interaction with technology be leveraged for automation and

improved user experience?

• How can user interaction data be leveraged for automated recommendations and im-

proved user experience in e-commerce?

• How can data from many different sources be used for improved product recommendations in e-commerce?


The researchers in the academic host environment already have significant experience of ad-

dressing many of these research questions, or related ones, in different contexts.

Senast uppdaterad av Susanne Lundborg