Event

Join the 23rd NLP Meetup on February 26th 2020 at the Geography Department of the University of Zurich. The topical landscape we are going to explore this time ranges from applications of NLP in Geography, over multi-aspect sentiment classification to the automatic construction of knowledge graphs.

Date & Time

26 February, 2020, 18:00 ~

Location (updated)

• Lecture Room: Y03-G-85 at University of Zurich, Irchel campus
• Networking: Hall Y24-G-100 "Lichthof" at University of Zurich, Irchel campus (link)

Registration

Registration is required to attend the event.

Please register on the meetup site: https://www.meetup.com/NLP-Zurich/events/268317918/

Event Agenda

18:00 ~ 18:10 Welcome

18:10 ~ 18:30 Talk I: Olga Koblet & Ross Purves: Doing Geography with Text

18:30 ~ 18:50 Talk II: Diego Antognini: Multi-Dimensional Explanation of Ratings from Reviews

18:50 ~ 19:10 Talk III: Janna Lipenkova: Automatic Learnings of Classifications in Highly Dynamic Domains

19:10 ~ Networking

Talk I: Doing Geography with Text

Abstract

Why would geographers be interested in working with text? We will show how natural language is an incredibly rich source of information for geographic research, introducing topics such as geographical information retrieval vernacular place names, exploring nearness and characterising landscape properties and perception.

Speakers

• Dr. Olga Koblet (Department of Geography, University of Zurich)

Olga Koblet's research is concerned with understanding and modelling ways in which people describe landscapes. She is particularly interested in automatic extraction of relevant information from written texts including travel reports, historical texts and descriptions of photographs contributed in crowdsourcing projects.

• Prof. Dr. Ross Purves (Department of Geography, University of Zurich)

Ross Purves is broadly interested in the use of text to explore geographically relevant research questions, and the development of methods to underpin such research, especially with respect to environmentally motivated research questions.

Talk II: Multi-Dimensional Explanation of Ratings from Reviews

Abstract

Automated predictions require explanations to be interpretable by humans. However, neural methods generally offer little transparency, and interpretability often comes at the cost of performance. In this paper, we consider explaining multi-aspect sentiments with text snippets from reviews, which suffice to make the prediction. Earlier work used attention mechanisms as a way of finding words that predict the sentiment towards a specific aspect and improving recommendation or summarization models. In our work, we propose a neural model that generates, in an unsupervised manner, probabilistic multi-dimensional masks that are interpretable and predict multi-aspect sentiment ratings.

Speaker

• Diego Antognini (Ph.D. candidate in Computer Science, Swiss Federal Institute of Technology in Lausanne (EPFL))

Diego Antognini is a second year Ph.D. candidate in Computer Science at Swiss Federal Institute of Technology in Lausanne (EPFL). He works in the Artificial Intelligence Laboratory (LIA) under the supervision of professor Boi Faltings in the areas of natural language processing, machine learning and game theory.

Talk III: Automatic Learnings of Classifications in Highly Dynamic Domains

Abstract

In this talk, I will summarize recent approaches to interpretable and linguistically motivated word embeddings. I will also discuss how these approaches can be used to enhance a model for the extraction of lexical relations, such as hypernymy and meronymy, with the ultimate goal of constructing knowledge graphs and hierarchical knowledge representations.

Speaker

• Dr. Janna Lipenkova (CEO of Anacode GmbH)

Dr. Janna Lipenkova is CEO of Anacode GmbH, a Berlin startup developing software for text analytics and market research in China. She holds an MA in Chinese Studies and Economics and a Ph.D. in Computational Linguistics (Freie Universität Berlin). Janna has a decade of working experience in text analytics for business use (Google, ABBYY Moscow and Lionbridge, among others) and has authored numerous publications and talks at international conferences.

Sponsors

Graduate Campus, University of Zurich

Geocomputation Unit, Department of Geography, University of Zurich

Associated Organizations

NLP Meetup Zurich, https://www.meetup.com/NLP-Zurich

Geocomputation Unit, Department of Geography, Univerzity of Zurich, https://www.geo.uzh.ch/en/units/gco.html

ML4SEP Peer Mentoring Group on Applied Machine Learning on Social & Environmental Problems, https://sites.google.com/view/ml4sep