Fix effects of reordering bug introduced in bulk metadata corrections from 2025-11-14

data/xml/2024.acl.xml

@@ -11132,9 +11132,9 @@
       <author><first>Rifki Afina</first><last>Putri</last><affiliation>Korea Advanced Institute of Science &amp; Technology</affiliation></author>
       <author><first>Emmanuel</first><last>Dave</last><affiliation>Binus University</affiliation></author>
       <author><first>Jhonson</first><last>Lee</last><affiliation>Tokopedia</affiliation></author>
-      <author><first>Nuur</first><last>Shadieq</last><affiliation>Binus University</affiliation></author>
-      <author><first>Wawan</first><last>Cenggoro</last><affiliation>Institut Teknologi Bandung</affiliation></author>
-      <author><first>Salsabil Maulana</first><last>Akbar</last><affiliation>Universitas Telkom</affiliation></author>
+      <author><first>Nuur</first><last>Shadieq</last><affiliation>Universitas Telkom</affiliation></author>
+      <author><first>Wawan</first><last>Cenggoro</last><affiliation>Binus University</affiliation></author>
+      <author><first>Salsabil Maulana</first><last>Akbar</last><affiliation>Institut Teknologi Bandung</affiliation></author>
       <author><first>Muhammad Ihza</first><last>Mahendra</last><affiliation>Universitas Telkom</affiliation></author>
       <author><first>Dea Annisayanti</first><last>Putri</last><affiliation>Universitas Indonesia</affiliation></author>
       <author><first>Bryan</first><last>Wilie</last><affiliation>Hong Kong University of Science and Technology</affiliation></author>

data/xml/2025.arabicnlp.xml

@@ -384,8 +384,8 @@
     <paper id="26">
       <title>Mind the Gap: A Review of <fixed-case>A</fixed-case>rabic Post-Training Datasets and Their Limitations</title>
       <author orcid="0009-0004-6842-1300"><first>Mohammed</first><last>Alkhowaiter</last><affiliation>Prince Sattam bin Abdulaziz University</affiliation></author>
-      <author><first>Norah</first><last>Alshahrani</last><affiliation>University of Bisha</affiliation></author>
-      <author><first>Saied</first><last>Alshahrani</last><affiliation>ASAS AI</affiliation></author>
+      <author><first>Norah</first><last>Alshahrani</last><affiliation>ASAS AI</affiliation></author>
+      <author><first>Saied</first><last>Alshahrani</last><affiliation>University of Bisha</affiliation></author>
       <author><first>Reem I.</first><last>Masoud</last></author>
       <author orcid="0000-0002-9914-915X"><first>Alaa</first><last>Alzahrani</last><affiliation>King Salman Global Academy for Arabic</affiliation></author>
       <author orcid="0000-0002-5082-8565"><first>Deema</first><last>Alnuhait</last><affiliation>University of Illinois at Urbana-Champaign</affiliation></author>
@@ -2103,11 +2103,11 @@
     </paper>
     <paper id="133">
       <title>Tokenizers United at <fixed-case>QIAS</fixed-case>-2025: <fixed-case>RAG</fixed-case>-Enhanced Question Answering for Islamic Studies by Integrating Semantic Retrieval with Generative Reasoning</title>
-      <author><first>Mohamed</first><last>Samy</last><affiliation>Institute</affiliation></author>
-      <author><first>Mayar</first><last>Boghdady</last></author>
+      <author><first>Mohamed</first><last>Samy</last></author>
+      <author><first>Mayar</first><last>Boghdady</last><affiliation>Institute</affiliation></author>
       <author><first>Marwan</first><last>El Adawi</last></author>
       <author><first>Mohamed</first><last>Nassar</last></author>
-      <author><first>Ensaf Hussein</first><last>Mohamed</last></author>
+      <author><first>Ensaf</first><last>Hussein</last></author>
       <pages>960-965</pages>
       <abstract/>
       <url hash="1572525d">2025.arabicnlp-sharedtasks.133</url>

data/xml/2025.emnlp.xml

@@ -21230,8 +21230,8 @@
       <author orcid="0000-0003-0049-4808"><first>Aloka</first><last>Fernando</last><affiliation>University of Moratuwa</affiliation></author>
       <author orcid="0000-0002-5361-4810"><first>Nisansa</first><last>de Silva</last><affiliation>University of Moratuwa</affiliation></author>
       <author><first>Menan</first><last>Velayuthan</last><affiliation>University of Moratuwa</affiliation></author>
-      <author orcid="0000-0003-0701-0204"><first>Charitha</first><last>Rathnayake</last><affiliation>Massey University</affiliation></author>
-      <author><first>Surangika</first><last>Ranathunga</last><affiliation>University of Moratuwa</affiliation></author>
+      <author><first>Charitha</first><last>Rathnayake</last><affiliation>University of Moratuwa</affiliation></author>
+      <author orcid="0000-0003-0701-0204"><first>Surangika</first><last>Ranathunga</last><affiliation>Massey University</affiliation></author>
       <pages>28252-28269</pages>
       <abstract>Parallel Data Curation (PDC) techniques aim to filter out noisy parallel sentences from web-mined corpora. Ranking sentence pairs using similarity scores on sentence embeddings derived from Pre-trained Multilingual Language Models (multiPLMs) is the most common PDC technique. However, previous research has shown that the choice of the multiPLM significantly impacts the quality of the filtered parallel corpus, and the Neural Machine Translation (NMT) models trained using such data show a disparity across multiPLMs. This paper shows that this disparity is due to different multiPLMs being biased towards certain types of sentence pairs, which are treated as noise from an NMT point of view. We show that such noisy parallel sentences can be removed to a certain extent by employing a series of heuristics. The NMT models, trained using the curated corpus, lead to producing better results while minimizing the disparity across multiPLMs. We publicly release the source code and the curated datasets</abstract>
       <url hash="7a30667d">2025.emnlp-main.1435</url>

data/xml/2025.starsem.xml

@@ -222,10 +222,10 @@
     <paper id="18">
       <title>Latent Traits and Cross-Task Transfer: Deconstructing Dataset Interactions in <fixed-case>LLM</fixed-case> Fine-tuning</title>
       <author><first>Shambhavi</first><last>Krishna</last><affiliation>University of Massachusetts at Amherst</affiliation></author>
-      <author orcid="0000-0003-4607-936X"><first>Atharva</first><last>Naik</last><affiliation>Department of Computer Science, University of Massachusetts at Amherst</affiliation></author>
+      <author><first>Atharva</first><last>Naik</last></author>
       <author><first>Chaitali</first><last>Agarwal</last></author>
       <author><first>Sudharshan</first><last>Govindan</last></author>
-      <author><first>Haw-Shiuan</first><last>Chang</last></author>
+      <author orcid="0000-0003-4607-936X"><first>Haw-Shiuan</first><last>Chang</last><affiliation>Department of Computer Science, University of Massachusetts at Amherst</affiliation></author>
       <author><first>Taesung</first><last>Lee</last></author>
       <pages>225-241</pages>
       <abstract>Large language models are increasingly deployed across diverse applications. This often includes tasks LLMs have not encountered during training.This implies that enumerating and obtaining the high-quality training data for all tasks is infeasible. Thus, we often need to rely on transfer learning using datasets with different characteristics, and anticipate out-of-distribution requests.Motivated by this practical need, we propose an analysis framework, building a transfer learning matrix and dimensionality reduction, to dissect these cross-task interactions.We train and analyze 10 models to identify latent abilities (e.g., Reasoning, Sentiment Classification, NLU, Arithmetic)and discover the side effects of the transfer learning.Our findings reveal that performance improvements often defy explanations based on surface-level dataset similarity or source data quality. Instead, hidden statistical factors of the source dataset, such as class distribution and generation length proclivities, alongside specific linguistic features, are actually more influential.This work offers insights into the complex dynamics of transfer learning, paving the way for more predictable and effective LLM adaptation.</abstract>

data/xml/2025.wmt.xml

@@ -1177,20 +1177,20 @@
       <author><first>Dinesh</first><last>Tewari</last><affiliation>Google</affiliation></author>
       <author><first>Baba Mamadi</first><last>Diane</last><affiliation>NKO USA INC</affiliation></author>
       <author><first>Djibrila</first><last>Diane</last><affiliation>NKO USA INC</affiliation></author>
-      <author><first>Solo Farabado</first><last>Cissé</last><affiliation>Stanford University</affiliation></author>
-      <author><first>Koulako Moussa</first><last>Doumbouya</last><affiliation>NKO USA INC</affiliation></author>
+      <author><first>Solo Farabado</first><last>Cissé</last><affiliation>NKO USA INC</affiliation></author>
+      <author><first>Koulako Moussa</first><last>Doumbouya</last><affiliation>Stanford University</affiliation></author>
       <author><first>Edoardo</first><last>Ferrante</last><affiliation>Conseggio pe-o patrimonio linguistico ligure</affiliation></author>
       <author><first>Alessandro</first><last>Guasoni</last><affiliation>Conseggio pe-o patrimonio linguistico ligure</affiliation></author>
-      <author><first>Christopher</first><last>Homan</last><affiliation>Paair Institute</affiliation></author>
-      <author><first>Mamadou K.</first><last>Keita</last><affiliation>NIT, Arunachal Pradesh</affiliation></author>
-      <author><first>Sudhamoy</first><last>DebBarma</last><affiliation>tyvan.ru</affiliation></author>
-      <author><first>Ali</first><last>Kuzhuget</last><affiliation>Stanford University</affiliation></author>
-      <author><first>David</first><last>Anugraha</last><affiliation>Universitas Indonesia</affiliation></author>
-      <author><first>Muhammad Ravi</first><last>Shulthan Habibi</last><affiliation>University of Zurich</affiliation></author>
-      <author><first>Sina</first><last>Ahmadi</last><affiliation>Google</affiliation></author>
-      <author><first>Anthony</first><last>Munthali</last><affiliation>Google</affiliation></author>
-      <author><first>Jonathan Mingfei</first><last>Liu</last></author>
-      <author><first>Jonathan</first><last>Eng</last></author>
+      <author><first>Christopher</first><last>Homan</last></author>
+      <author><first>Mamadou K.</first><last>Keita</last><affiliation>Paair Institute</affiliation></author>
+      <author><first>Sudhamoy</first><last>DebBarma</last><affiliation>NIT, Arunachal Pradesh</affiliation></author>
+      <author><first>Ali</first><last>Kuzhuget</last><affiliation>tyvan.ru</affiliation></author>
+      <author><first>David</first><last>Anugraha</last><affiliation>Stanford University</affiliation></author>
+      <author><first>Muhammad Ravi</first><last>Shulthan Habibi</last><affiliation>Universitas Indonesia</affiliation></author>
+      <author><first>Sina</first><last>Ahmadi</last><affiliation>University of Zurich</affiliation></author>
+      <author><first>Anthony</first><last>Munthali</last></author>
+      <author><first>Jonathan Mingfei</first><last>Liu</last><affiliation>Google</affiliation></author>
+      <author><first>Jonathan</first><last>Eng</last><affiliation>Google</affiliation></author>
       <pages>1103-1123</pages>
       <abstract>We open-source SMOL (Set of Maximal Over-all Leverage), a suite of training data to un-lock machine translation for low-resource languages (LRLs). SMOL has been translated into123 under-resourced languages (125 language pairs), including many for which there exist no previous public resources, for a total of 6.1M translated tokens. SMOL comprises two sub-datasets, each carefully chosen for maximum impact given its size: SMOLSENT, a set of sentences chosen for broad unique token coverage, and SMOLDOC, a document-level source focusing on a broad topic coverage. They join the already released GATITOS for a trifecta of paragraph, sentence, and token-level content. We demonstrate that using SMOL to prompt or fine-tune Large Language Models yields robust chrF improvements. In addition to translation, we provide factuality ratings and rationales for all documents in SMOLDOC, yielding the first factuality datasets for most of these languages.</abstract>
       <url hash="b069adea">2025.wmt-1.85</url>