NLP Master project entitled „Sense Clustering Over Time" from LT group won EXPO 2019
17 October 2019, by Seid Muhie Yimam
Congratulations to Inga Kempfert!!
Inga Kempfert, who has participated in the "Master project Web Interfaces for Language Processing Systems" of the summer semester 2019, has won the EXPO 2019 (https://www.inf.uni-hamburg.de/inst/basecamp/blog/19-07-24-expo2019.html).
Title of the project: SCOT - Sense Clustering Over Time
Short Description: SCoT SenseViewer is a web application that allows users to view the change of word meanings. Data-driven means that there is a large amount of language data behind the tool, the relationships of which are visualized through the graphical user interface.
Humanities scholars can use the tool to better investigate questions in the field of historical semantics. For example, the change in meaning of often polysemous basic concepts such as "freedom", "power", "life", "crisis" can be traced. The respective meanings give us information about the cultural, political and scientific descriptions of the world and self-descriptions of the respective time.
The meanings of a word are presented as a clustered graph. Each cluster represents meaning and is color-coded. All nodes in the graph occur in the data in a similar context as the target word. This means that they have a similar meaning as the target word. If the nodes among themselves also have a semantic similarity, they are connected with an edge.
Users can edit the computed clusters, and look at each time slice in the graph individually to get an idea of how the meaning has changed.