We are delighted to announce that Malte Ostendorff has successfully defended his PhD thesis titled ‘‘Aspect-based Document Similarity for Literature Recommender Systems’’ on April 25th, 2023.
The research objective of Malte’s thesis was to improve literature recommendation systems by proposing and implementing a novel concept termed ‘aspect-based document similarity’, which can better capture the multifaceted components of document similarity. His research highlights how two documents may be similar in nuanced ways, for example, by using a similar methodology, addressing a similar task, or using the same dataset, despite covering different research problems.
The thesis makes significant contributions in three areas: firstly, by demonstrating the need for aspect information in document similarity measures; secondly, by introducing SciNCL, a new document representation method that surpasses the current state-of-the-art; and thirdly, by crafting two practical approaches for aspect-based document similarity that overcome the limitations of existing aspect-free similarity measures.
Malte plans to remain in close collaboration with our group as he begins his postdoc research.
- The prototype described in the thesis can be explored here: https://recsys.ostendorff.org/
- The key research paper that presents his PhD work is available here: https://www.gipp.com/wp-content/papercite-data/pdf/ostendorff2020c.pdf
- View all of Malte’s publications here
- Find Malte Ostendorff on Twitter