Abstract

This paper describes our system used in the SemEval-2023 Task 11 Learning With Dis- agreements (Le-Wi-Di). This is a subjective task since it deals with detecting hate speech, misogyny and offensive language. Thus, dis- agreement among annotators is expected. We experiment with different settings like loss functions specific for subjective tasks and in- clude anonymized annotator-specific informa- tion to help us understand the level of disagree- ment. We perform an in-depth analysis of the performance discrepancy of these differ- ent modelling choices. Our system achieves a cross-entropy of 0.58, 4.01 and 3.70 on the test sets of HS-Brexit, ArMIS and MD-Agreement, respectively. Our code implementation is pub- licly available.

Abstract

Identifying human values behind arguments is a complex task which requires understanding of premise, stance and conclusion together. We propose a method that uses a pre-trained lan- guage model, DeBERTa, to tokenize and con- catenate the text before feeding it into a fully connected neural network. We also show that leveraging the hierarchy in values improves the performance by .14 F1 score compared to only using level 2 values. Our code is made publicly available here.1

Abstract

This paper describes our approach for SemEval- 2023 Task 10: Explainable Detection of Online Sexism (EDOS). The task deals with identifi- cation and categorization of sexist content into fine-grained categories for explainability in sex- ism classification. The explainable categoriza- tion is proposed through a set of three hierar- chical tasks that constitute a taxonomy of sexist content, each task being more granular than the former for categorization of the content. Our team (iREL) participated in all three hierarchi- cal subtasks. Considering the inter-connected task structure, we study multilevel training to study the transfer learning from coarser to finer tasks. Our experiments based on pretrained transformer architectures also make use of ad- ditional strategies such as domain-adaptive pre- training to adapt our models to the nature of the content dealt with, and use of the focal loss objective for handling class imbalances. Our best-performing systems on the three tasks achieve macro-F1 scores of 85.93, 69.96 and 54.62 on their respective validation sets

Abstract

With the advent of multilingual models like mBART, mT5, IndicBART etc., summarization in low resource Indian languages is getting a lot of attention now a days. But still the number of datasets is low in number. In this work, we (Team HakunaMatata) study how these multilingual models perform on the datasets which have Indian languages as source and target text while performing summarization. We experimented with IndicBART and mT5 models to perform the experiments and report the ROUGE-1, ROUGE-2, ROUGE-3 and ROUGE-4 scores as a performance metric.

Abstract

Lack of encyclopedic text contributors, especially on Wikipedia, makes automated text generation for low resource (LR) languages a critical problem. Existing work on Wikipedia text generation has focused on English only where English reference articles are sum- marized to generate English Wikipedia pages. But, for low-resource languages, the scarcity of reference articles makes monolingual summarization inefective in solving this problem. Hence, in this work, we propose XWikiGen, which is the task of cross-lingual multi-document summarization of text from multiple reference ar- ticles, written in various languages, to generate Wikipedia-style text. Accordingly, we contribute a benchmark dataset, XWikiRef, spanning ∼69K Wikipedia articles covering fve domains and eight languages. We harness this dataset to train a two-stage system where the input is a set of citations and a section title and the output is a section-specifc LR summary. The proposed system is based on a novel idea of neural unsupervised extractive summariza- tion to coarsely identify salient information followed by a neural abstractive model to generate the section-specifc text. Extensive experiments show that multi-domain training is better than the multi-lingual setup on average. We make our code and dataset publicly availableLack of encyclopedic text contributors, especially on Wikipedia, makes automated text generation for low resource (LR) languages a critical problem. Existing work on Wikipedia text generation has focused on English only where English reference articles are sum- marized to generate English Wikipedia pages. But, for low-resource languages, the scarcity of reference articles makes monolingual summarization inefective in solving this problem. Hence, in this work, we propose XWikiGen, which is the task of cross-lingual multi-document summarization of text from multiple reference ar- ticles, written in various languages, to generate Wikipedia-style text. Accordingly, we contribute a benchmark dataset, XWikiRef, spanning ∼69K Wikipedia articles covering fve domains and eight languages. We harness this dataset to train a two-stage system where the input is a set of citations and a section title and the output is a section-specifc LR summary. The proposed system is based on a novel idea of neural unsupervised extractive summariza- tion to coarsely identify salient information followed by a neural abstractive model to generate the section-specifc text. Extensive experiments show that multi-domain training is better than the multi-lingual setup on average. We make our code and dataset publicly available

Abstract

Named Entity Recognition(NER) is a task of recognizing entities at a token level in a sen- tence. This paper focuses on solving NER tasks in a multilingual setting for complex named entities.Our team, LLM-RM partici- pated in the recently organized SemEval 2023 task, Task 2: MultiCoNER II,Multilingual Complex Named Entity Recognition. We approach the problem by leveraging cross- lingual representation provided by fine-tuning XLM-Roberta base model on datasets of all of the 12 languages provided - Bangla, Chinese, English, Farsi, French, German, Hindi, Italian, Portuguese, Spanish, Swedish and Ukrainian

Abstract

Factual consistency is one of the most important requirements when editing high quality documents. It is extremely important for automatic text generation systems like summarization, question answering, dialog modeling, and language modeling. Still, automated factual inconsistency detection is rather under-studied. Existing work has focused on (a) finding fake news keeping a knowledge base in context, or (b) detecting broad contradiction (as part of natural language inference literature). However, there has been no work on detecting and explaining types of factual inconsistencies in text, without any knowledge base in context. In this paper, we leverage existing work in linguistics to formally define five types of factual inconsistencies. Based on this categorization, we contribute a novel dataset, FICLE (Factual Inconsistency CLassification with Explanation), with ~8K samples where each sample consists of two sentences (claim and context) annotated with type and span of inconsistency. When the inconsistency relates to an entity type, it is labeled as well at two levels (coarse and fine-grained). Further, we leverage this dataset to train a pipeline of four neural models to predict inconsistency type with explanations, given a (claim, context) sentence pair. Explanations include inconsistent claim fact triple, inconsistent context span, inconsistent claim component, coarse and fine-grained inconsistent entity types. The proposed system first predicts inconsistent spans from claim and context; and then uses them to predict inconsistency types and inconsistent entity types (when inconsistency is due to entities). We experiment with multiple Transformer …

Abstract

This paper describes our system (iREL) for Tweet intimacy analysis shared task of the SemEval 2023 workshop at ACL 2023. Our system achieved an overall Pearson’s r score of 0.5924 and ranked 10th on the overall leaderboard. For the unseen languages, we ranked third on the leaderboard and achieved a Pearson’s r score of 0.485. We used a single multilingual model for all languages, as discussed in this paper. We provide a detailed description of our pipeline along with multiple ablation experiments to further analyse each component of the pipeline. We demonstrate how translation-based augmentation, domainspecific features, and domain-adapted pretrained models improve the understanding of intimacy in tweets. The code can be found at https://github.com/bhavyajeet/Multilingualtweet-intimacy

Abstract

Fact-to-text generation can allow for the generation of high-quality, informative texts such as Wikipedia articles. Cross-lingual fact-to-text generation (XF2T) involves using facts available in a language, typically English, and generating texts in a different language based on these facts. This is particularly relevant for low and medium-resource languages, which have relatively structured informative content. This work explores the problem of XF2T for generating long text from given facts with a specific focus on generating factually grounded content. Unfortunately, previous work either focuses on cross-lingual facts to short text or monolingual graph to text generation. In this paper, we propose a novel solution to the multi-sentence XF2T task, which addresses these challenges by training multilingual Transformer-based models with coverage prompts and rebalanced beam search, and further improving the quality by defining task-specific reward functions and training on them using reinforcement learning. Keywords: XF2T, text generation, cross-lingual, NLG evaluation, low resource NLG

Abstract

Augmenting Wikidata with spatial relations specific to Geography can be useful for increasing its utility in multiple applications. In this paper, we aim to extract orientation of borders between countries in the world, from their Wikipedia text and suggest its use to augment the shares_borders_with relation in Wikidata. We propose the use of Natural Language Inference (NLI) for extracting the orientation relations from text and show that when combined with contextual lexical patterns, the performance becomes better than the standard NLI setting.