@phdthesis{Renker2015,
  type      = {Master Thesis},
  author    = {Lennart Renker},
  title     = {Exploration von Textkorpora - Topic Models als Grundlage der Interaktion},
  url       = {https://nbn-resolving.org/urn:nbn:de:hbz:832-epub4-6686},
  pages     = {117},
  year      = {2015},
  abstract  = {The internet offers a seemingly infinite amount of information. A central issue of our days is to make use of this nformation. To formulate efficient search queries a user must have good domain knowledge. Often this is not the case, wherefore a lot of time has to be invested to get an overview of the topic in question. In those situations a user ends up in an exploratory search process, in which he has to outline the individual topics step by step. By now machine learning algorithms are frequently used in data organization but stay invisible to the user for most of the time. The interactive use of these methods could optimize search processes by connecting the human ability to judge with machine processing powers used on large data sources. Topic models are algorithms that find latent topics in unstructured text corpora and are relatively good to interpret. Their use is promising in exploratory search processes, in which a user has to gain new domain knowledge quickly. It appears that many researches use topic models mostly to generate static visualizations of latent text structures. Sensemaking is an essential part of exploratory search processes but it is still only used to a small extent in order to justify algorithmic novelties and bring them into a larger context. Therefore the assumption is derived, that the use of sensemaking models and user centered concepts for exploratory search processes could lead to the development of new ways to interact with topic models and to generate a framework for publications of the correlating research fields.},
  language  = {de}
}