no code implementations • NoDaLiDa 2021 • Lovisa Hagström, Richard Johansson
The field encompasses different methods that aim to preserve the performance of a model while decreasing the size of it.
1 code implementation • CLASP 2022 • Lovisa Hagström, Tobias Norlund, Richard Johansson
This is a setting in which we fuse language with information from the math modality and strive to replicate some fusion methods from the vision-and-language domain.
no code implementations • WS (NoDaLiDa) 2019 • Joakim Åkerström, Adel Daoud, Richard Johansson
Social science researchers often use text as the raw data in investigations: for instance, when investigating the effects of IMF policies on the development of countries under IMF programs, researchers typically encode structured descriptions of the programs using a time-consuming manual effort.
no code implementations • 25 Mar 2024 • Nicolas Audinet de Pieuchon, Adel Daoud, Connor Thomas Jerzak, Moa Johansson, Richard Johansson
We investigate the potential of large language models (LLMs) to disentangle text variables--to remove the textual traces of an undesired forbidden variable in a task sometimes known as text distillation and closely related to the fairness in AI and causal inference literature.
no code implementations • 24 Mar 2024 • Richard Johansson
We investigate the behavior of methods that use linear projections to remove information about a concept from a language representation, and we consider the question of what happens to a dataset transformed by such a method.
1 code implementation • 2 Nov 2023 • Lovisa Hagström, Denitsa Saynova, Tobias Norlund, Moa Johansson, Richard Johansson
In this work, we identify potential causes of inconsistency and evaluate the effectiveness of two mitigation strategies: up-scaling and augmenting the LM with a retrieval corpus.
1 code implementation • 25 May 2023 • Ehsan Doostmohammadi, Tobias Norlund, Marco Kuhlmann, Richard Johansson
Inspired by this, we replace the semantic retrieval in Retro with a surface-level method based on BM25, obtaining a significant reduction in perplexity.
1 code implementation • 17 Apr 2023 • Mehrdad Farahani, Richard Johansson
Autoregressive models used to generate responses in open-domain dialogue systems often struggle to take long-term context into account and to maintain consistency over a dialogue.
no code implementations • 23 Feb 2023 • Tobias Norlund, Ehsan Doostmohammadi, Richard Johansson, Marco Kuhlmann
Recent work on the Retrieval-Enhanced Transformer (RETRO) model has shown that off-loading memory from trainable weights to a retrieval database can significantly improve language modeling and match the performance of non-retrieval models that are an order of magnitude larger in size.
no code implementations • 3 Feb 2023 • Manuj Malik, Richard Johansson
We propose a methodology and design two benchmark sets for measuring to what extent language-and-vision language models use the visual signal in the presence or absence of stereotypes.
1 code implementation • COLING 2022 • Lovisa Hagström, Richard Johansson
To find the best approach, we investigate and compare seven possible methods for adapting three different pre-trained VL models to text-only input.
1 code implementation • ACL 2022 • Lovisa Hagström, Richard Johansson
Primarily, we find that the visual commonsense knowledge is not significantly different between the multimodal models and unimodal baseline models trained on visual text data.
no code implementations • NAACL 2022 • Adel Daoud, Connor T. Jerzak, Richard Johansson
However, these methods rely on a critical assumption that there is no treatment leakage: that is, the text only contains information about the confounder and no information about treatment assignment.
no code implementations • EMNLP (BlackboxNLP) 2021 • Tobias Norlund, Lovisa Hagström, Richard Johansson
We find that our method can successfully be used to measure visual knowledge transfer capabilities in models and that our novel model architecture shows promising results for leveraging multimodal knowledge in a unimodal setting.
no code implementations • LREC 2020 • Richard Johansson, Yvonne Adesam
We investigate a transition-based parser that uses Eukalyptus, a function-tagged constituent treebank for Swedish which includes discontinuous constituents.
no code implementations • LREC 2020 • Kathrein Abu Kwaik, Stergios Chatzikyriakidis, Simon Dobnik, Motaz Saad, Richard Johansson
As the number of social media users increases, they express their thoughts, needs, socialise and publish their opinions reviews.
no code implementations • CONLL 2018 • Murhaf Fares, Stephan Oepen, Lilja {\O}vrelid, Jari Bj{\"o}rne, Richard Johansson
We summarize empirical results and tentative conclusions from the Second Extrinsic Parser Evaluation Initiative (EPE 2018).
no code implementations • COLING 2018 • Luis Nieto-Pi{\~n}a, Richard Johansson
Automatically learnt word sense embeddings are developed as an attempt to refine the capabilities of coarse word embeddings.
no code implementations • IJCNLP 2017 • Luis Nieto-Pi{\~n}a, Richard Johansson
We propose to improve word sense embeddings by enriching an automatic corpus-based method with lexicographic data.
no code implementations • WS 2017 • Olof Mogren, Richard Johansson
We also show that using the auxiliary task of learning the relation type speeds up convergence and improves the prediction accuracy for the word generation task.
no code implementations • COLING 2016 • Anna Ehrlemark, Richard Johansson, Benjamin Lyngfelt
Finding authentic examples of grammatical constructions is central in constructionist approaches to linguistics, language processing, and second language learning.
no code implementations • WS 2016 • Wafia Adouane, Nasredine Semmar, Richard Johansson
The ALI standard methods require datasets for training and use character/word-based n-gram models.
no code implementations • WS 2016 • Wafia Adouane, Nasredine Semmar, Richard Johansson, Victoria Bobicev
Automatic Language Identification (ALI) is the detection of the natural language of an input text by a machine.
no code implementations • WS 2016 • Wafia Adouane, Nasredine Semmar, Richard Johansson
In sub-task 2, which deals with Arabic dialect identification, the system achieved its best performance using character-based n-grams (49. 67{\%} accuracy), ranking fourth in the closed track (the best result being 51. 16{\%}), and an accuracy of 53. 18{\%}, ranking first in the open track.
no code implementations • LREC 2016 • Richard Johansson, Yvonne Adesam, Gerlof Bouma, Karin Hedberg
We describe the word sense annotation layer in \textit{Eukalyptus}, a freely available five-domain corpus of contemporary Swedish with several annotation layers.
no code implementations • LREC 2016 • Wafia Adouane, Richard Johansson
To fill this gap, we created these two main linguistic resources.
no code implementations • RANLP 2015 • Luis Nieto Piña, Richard Johansson
Distributed representations of words have boosted the performance of many Natural Language Processing tasks.
no code implementations • LREC 2012 • Richard Johansson, Karin Friberg Heppin, Dimitrios Kokkinakis
We present the first results on semantic role labeling using the Swedish FrameNet, which is a lexical resource currently in development.
no code implementations • LREC 2012 • Sucheta Ghosh, Richard Johansson, Giuseppe Riccardi, Sara Tonelli
We describe two constraint-based methods that can be used to improve the recall of a shallow discourse parser based on conditional random field chunking.