Analytical Information Systems
Migraine Radar (MiRa)
The project „Migraine Radar“ is to examine to which extent text data of Web 2.0 communication platforms can serve as basis for studies, in particular of medical studies. The idea of the project is to provide an exemplary study of the connection between more migraine attacks and extremely changing weather conditions. For this purpose, we analyze news from the microblogging platform Twitter about migraine attacks and afterwards these ‘tweets’ are linked to weather data of the Deutscher Wetterdienst (German Weather Service). The connection of datasets is made by using the information on the location of the tweeters. Some undergraduate assistants and two students who complete their Bachelor's thesis in the framework of this project are currently involved in this project.
For more information, please visit www.migraene-radar.de.
The project MiRa is one of the winners of the competition “What makes healthy?” by the Federal Ministry of Education and Research. When the winners were announced, Annette Schavan, Federal Minister of Education and Research, said, "The exciting and versatile project ideas show in an impressive way that Germany has excellent and committed talents in the field of health research.”
politwi - analysis of political tweets for the parliamentary election in 2013
politwi consists of politics and Twitter. The aim is to analyze freely available political tweets ahead of the parliamentary elections 2013 in Germany. Here, the used hashtags are analyzed and identified in particular the latest trends. It is a research project at the Institute for Information Systems (iisys) at the Hof University in collaboration with the "Big Data Analytics Research Lab" at the Goethe University Frankfurt.
For more information, please visit www.politwi.de.
Opinion mining for the insurance industry
This project aims to mechanically evaluate expressed opinions in Web 2.0 platforms which refer to insurance companies and their products or services. It is examined how opinion mining can be optimized for the German language so that it can be used profitably by the insurance sector. The project is carried out in close co-operation with nobiscum GmbH and promoted by the Bavarian State Ministry of Economics, Infrastructure, Transport and Technology.
Opinionlist I - Generating a list of opinion-forming adjectives for the German language
Opinion mining extracts opinions from unstructured texts. Opinions are often expressed by using adjectives (e.g. "This mobile is a bad one.”) The aim of the project is to match numerical values with those opinion-forming adjectives which show how positive or negative an expressed opinion is. For this purpose, we use a method of machine learning which is based on the analysis of product reviews.
Opinionlist II - Generating a list of opinion-forming nouns for the German language
The aim of this project is to create a list of nouns together with values regarding opinions. For example, a negative value would be assigned to the word “waste” whereas “success” would get a positive value. The list is based on customer reviews and lexical relations (e.g. synonyms, antonyms). The opinion list can be used for future projects in the field of opinion mining, e.g. for the determination of the general sentiment (positive or negative) of a text.