Book Search and Academic Search

This edition of SEA is in partnership with FNWI (http://www.uva.nl/en/about-the-uva/organisation/faculties/content/faculteit-der-natuurwetenschappen-wiskunde-en-informatica/faculty-of-science.html). We will have two talks followed by drinks. Marijn Koolen (http://humanities.uva.nl/~mkoolen1/) (formerly at UvA) from Huygens ING (https://www.huygens.knaw.nl/?lang=en) will give the academic talk. The industry talk will be given by Marius Doornenbal from Elsevier (https://www.elsevier.com). Title and abstract of the industry talk will be announced later.

Please note that this edition of SEA will be held in SPUI25.

Program:

16:00 - 16:30 Marijn Koolen

16:30 - 17:00 Marius Doornenbal

17:00 - 18:00 Drinks & Snacks

Details of the talks:

-----------------------------------------------------------------------------------

Marijn Koolen--Exploiting user interactions to support complex book search tasks
Book-related social media provide rich information on how readers think about books, what aspects they care about, and how people search for and recommend books. In this talk I first discuss how book search information needs in online book discussions represent complex tasks that are not well supported by search and recommender systems, such as finding books in a particular style, books that complement a list of known books and finding a good reading order. Then I'll argue how exploiting the structure of book-related user interactions in different ways can support such complex search tasks.

Marijn Koolen has been involved in Information Retrieval (IR) research since 2005, working on aspects of link analysis and diversity in web search, cultural heritage IR and XML retrieval. His current research is on book search in the context of social media and IR systems to support complex search tasks. He got his Ph.D. at the University of Amsterdam and held positions as assistant professor of Information Science and Digital Humanities. He is currently working as a researcher and developer at Huygens ING in the Netherlands.

-----------------------------------------------------------------------------------

Marius Doornenbal--Topic Modelling a Science Domain

At Elsevier we have developed a system that scans full text literature as present in our data repositories, and extracts relevant and targeted pieces of textual information around certain topics. Starting out from the domain of neuroscience, we branch out the solution to other domains such as chemistry and other life sciences. The solution identifies the mention of specific topics in the text, and then goes on to qualify the text containing the concept mention – is this a definition of the concept? Is the concept associated with another important nugget of information – background, methodology, experimental results? On the Science Direct pages where end-users read articles, we can surface this information as background reading in the sidebar, and help readers navigate to Topic Pages that provide summarizations of relevant back-ground reading articles around specific topics of interest.

Marius Doornenbal is an NLP Scientist at Elsevier. His main interest is in Information Extraction and –generally– making knowledge available to different audiences – researchers, clinicians, corporate research labs , Elsevier is a world-leading provider of information solutions that enhance the performance of science, health, and technology professionals, empowering them to make better decisions, deliver better care, and sometimes make groundbreaking discoveries that advance the boundaries of knowledge and human progress. Elsevier provides web-based, digital solutions and publishes a large number of journals and book titles. Marius Doornenbal is part of the Content and Innovation group in Elsevier, building and managing the information extraction pipelines that power the information solutions serving specific corporate and research audiences. Marius Doornenbal has a PhD in Linguistics from the University of Leiden and published a few articles on his work in Elsevier.