HWS 2016 - FSS 2017
- Daily Log: Healthcare management for individuals with a mobile application
HWS 2015 - FSS 2016
- Prediction of daily routine of individuals for Healthcare management
- Predictive Analytics for Fresh Food Demand
- Building a Lexicon of the Financial Domain
HWS 2014 - FSS 2015
- Personal Process Guidance for Healthcare Management
- Relation Discovery for Open Information Extraction
- Data Stream Mining for Predictive Mainenance
- DBpedia++: Extending the Knowledge in DBpedia from Diverse Sources
FSS 2014 - HWS 2014
- Building a recommender system for Wiley's Online Library
HWS2013 - FSS2014
- Large-Scale Information Extraction from the Web
Daily Log: Healthcare management for individuals with a mobile application
The development of wearable devices such as smart-phones features a variety of sensors and provides new opportunities for continuous monitoring and supporting. In this project, we focus on diabetes patients where certain complications can be avoided by lifestyle changes. The goal is to identify and interpret common activity patterns but also which kind of activities are mandatory. In this context, we want to develop an mobile application that provides these functionalities.
Prediction of the Daily Routine of Individuals for Healthcare Management
Diabetes is a metabolic disease with high blood sugar levels over a prolonged period. In order to avoid critical long term effects, Diabetes patients have to adhere to a rather strict daily routines with respect to physical activity, food consumption and medication. In this project, we want to build a mobile system that makes predictions and recommandations based on the daily routine to support diabetes patients. In this context, the relevant activities and their relations must be identified and modeled. An available Android App enables to record the daily routine by hand and allows to collect context related information such as the location or the acceleration of the mobile device.
Predictive Analytics for Fresh Food Demand
Personal Process Guidance for Healthcare Management
Diabetes is a metabolic disease with high blood sugar levels over a prolonged period. In order to avoid critical long term effects, Diabetes patients have to adhere to a rather strict daily routines with respect to physical activity, food consumption and medication. In this project, we want to build a mobile App that analyzes the activities and daily routines of patients based on sensor data from the Smart Phone and provides guidance and support for a healthy lifestyle.
Data Stream Mining for Predictive Mainenance
Predictive Maintenance is a new paradigm for maintaining technical systems where the status of the system is constantly monitored and the resulting sensor data is analyzed for patterns that gives hint towards a future failure of the system. In this project, we analyze operational data from Printing machines provided by Heidelberger Druckmaschinen AG. We apply state of the Art sequence mining methods on huge amounts of log data in order to find new meaningful patterns for improving the online monitoring of the rpinting machines Worldwide.
DBpedia++: Extending the Knowledge in DBpedia from Diverse Sources
DBpedia is a large, multi-lingual, cross-domain knowledge base in the Semantic Web, extracted from Wikipedia. In this team project, different strategies for improving DBpedia are explored. In particular, various external data sources are planned to be used for enriching DBpedia.
Building a recommender system for Wiley's Online Library
Recommender systems help users to filter information by recommending items, in this case journal articles to users. To this end, a system exploits the information about past interactions of this user, but also of other users, with the library. The aim for this Master team project is to build a recommendation system for the article recommendations for Wiley-VCH's library based on a large, real-world dataset from Wiley.
Large-Scale Information Extraction from the Web
In this project, web tables are exploited as a source for knowledge base construction at large scale. Starting from the Common Crawl corpus, a pipeline of multiple steps is run:
- Separation of content and non-content (e.g., layout) tables in HTML pages
- Normalization of tables
- Row and column header detection
- Data extraction from tables
On the use case of country data, a large knowledge base with data about countries is created. One central outcome of the team project was the Web Data Commons web tables corpus.
- Dr. Heiko Paulheim