Between May 2 and May 16, there will be a home exam in the form of a ‘baby’ research project. To pass the exam, students need to submit and have accepted a scientific paper to the First IN5550 Workshop on Neural Natural Language Processing (WNNLP 2019).
Final examination in this class takes the form of a ‘home exam’, i.e. a project that students can work on over a period of two weeks. Like for the exam-qualifying obligatory assignments, group work (in teams of up to three students) is encouraged. Team composition needs to be declared before the start of the exam period and, thus, cannot be changed after May 1, 2019. Also, each team needs to decide beforehand which of the four available tracks they want to research: (1) Named Entity Recognition; (2) Natural Language Inference; (3) Negation Scope Resolution; or (4) Relation Extraction. Please announce your team composition and choice of track no later than May 1 by emailing the course contact address email@example.com.
The exam period will be Thursday, May 2, to Thursday, May 16, 2019. On the first day of the exam period, each team will receive a written set of suggestions for how to approach the track they have chose. During the exam period, there will be no lectures, but group sessions will run according to the regular schedule. On May 9, each project team will receive an individual mentoring session (of up to 30 minutes) with one or two of the track chairs (the course instructors). Upon submission of the home exam (in the form of a scientific paper), there will be a peer reviewing period where we also require students to contribute feedback on submissions by others (in a double-blind, anonymous fashion); reviews will be due on or before May 25. Each team will be required to revise their manuscript according to the feedback from reviewers and the area chairs and submit the final, camera-ready version by June 2; furthermore, each team will be (obligatorily) invited to give a short presentation of their approach and results in the laboratory slot on June 11, which is the final gathering of the class this term.
|Start of Review Period||Tuesday, May 21|
|End of Review Period||Saturday, May 25|
|Track Chair Decisions||Monday, May 27|
|Camera-Ready Manuscripts||Sunday, June 2|
|Oral Presentations||Tuesday, June 11|
Submissions must be formatted according to the standard style files of the Annual Meeting of the Association for Computational Linguistics (ACL 2019): Please start from the ACL template files for the easy-to-use and premium-quality LaTeX typesetting system (strongly recommended) or M$ Word. The laboratory session on April 30, 2019, will be devoted to getting started with paper writing in LaTex. Submissions must not exceed nine pages in total length, including all tables and figures, but not counting the list of bibliographic references (i.e. the final, unnumbered References section). All papers submitted for review must be anonymous, i.e. not reveal author identities. However, for full replicability, each submission must be backed up by a private repository (in the UiO installation of Microsoft GitHub, as always) providing all code, scripts, possibly additional data, and at least minimal instructions on how to run and evaluate the system. The code repository will only be visible to track chairs, i.e. please make sure to share it with the instructors in Microsoft GitHub. All submission must be uploaded electronically no later than May 16 to the on-line conference management site; it will be possible to revise and resubmit, so please make your initial submission well before the deadline.
Nine submissions have been accepted for oral presentation during the workshop (June 11, 2019; Seminar room Perl):
|14:15–14:20||Programme Chairs||Welcome and Workshop Overview|
|14:20–14:30||Christian Finnstrom, Sigbjørn Aukland and Åsmund Hugo||RNN for Named Entity Recognition: A Study of Design Strategies|
|14:30–14:40||Egil Ronningstad||Norwegian Named Entity Recognition|
|14:40–14:50||Erik Winge||Named Entity Recognition with Norwegian Text|
|14:50–15:00||Alexandra Huynh||Reading Science: An Analysis on Relation Classification|
|15:00–15:10||Eirik Magnussen||Negation Scope Resolution|
|15:10–15:20||Chris Ghai||Negation Scope Resolution: Quantifying Neural Uncertainty In An Imbalanced Setting|
|15:20–15:30||Skjalg Eriksen||Spring 2019 Home Exam: Natural Language Inference Encoders|
|15:30–15:40||Idir Berg Ould-Saada and Aleksander Bai||A Sensitivity Analysis of Recurrent Neural Networks in Natural Language Inference|
|15:40–15:50||Kjersti Viksaas, Anders Næss Evensen, and Inger Nordby Grønn||Nothing Compares 2 (Bi)LSTM|
|15:50–16:30||Everyone||Award Ceremony and Social Gathering|