Recommender Systems & Extraction Services

Extraction and recommender systems have evolved into a universal tool that support personal information management and help to explore large knowledge respositories.

Our research topics:

  • Text Mining, Entity Extraction and Disambiguation
  • Automatic Text Summarization
  • Semi-automatic Concept Alignment
  • Context-based recommendation awareness
  • Social Computing & Machine Learning

Recommender systems (also known as recommendation systems, recommendation engines, recommendation frameworks, recommendation platforms or simply recommender) work from a specific type of information filtering system technique that attempts to recommend information items (movies, TV program/show/episode, video on demand, music, books, news, images, web pages, scientific literature such as research papers etc.) or social elements (e.g. people, events or groups) that are likely to be of interest to the user.

These recommendations can either be triggered by the monitoring of personal information habits (i.e. collaborative filtering as provided by Amazon.com) or by comparing digital objects and their semantic similarity (i.e. similarity search as of Google Recipes) to each other. While methods that are built on information consumption patterns usually tend to impose privacy sensitive issues to the user and the corresponding service provider, the second approach is generally privacy unsensitive and ethically unproblematic. Nevertheless recent developments as practiced by recommendation service providers tend to combine both techniques in improving the recommendation quality and hence leveraging the user experience.

Our current research in the area of recommendation services focusses on the second approach. By calculating and comparing the "semantic imprint" of digital objects to each other, measures can be derived that help to identify relevant objects based on their semantic characteristics.