updates.json

[
{"title": "TTLab Paper Accepted at ArabicNLP 2026 Daleel Shared Task", "content": "We are happy to announce that our paper 'STAR-Ar: Sequence Tagging for Argument Recognition in Arabic' by Bhuvanesh Verma, Ali Abusaleh, and Alexander Mehler has been accepted at the SIGARAB ArabicNLP 2026 Daleel Shared Task. The paper presents a sequence-tagging approach to argument component identification and classification in Arabic text.", "date": "2026-09-14", "author": "Bhuvanesh Verma", "email": "verma@em.uni-frankfurt.de"},
{"title": "TTLab Paper Accepted at ArabicNLP 2026 AlexandriaX Shared Task", "content": "We are happy to announce that our paper 'A Fine-Tuned Surface Tagger for Arabic Machine-Translation Error-Span Detection and Classification' by Ali Abusaleh, Bhuvanesh Verma, and Alexander Mehler has been accepted at the SIGARAB ArabicNLP 2026 AlexandriaX Shared Task. The work addresses error-span detection and classification for dialectal Arabic machine translation.", "date": "2026-09-14", "author": "Ali Abusaleh", "email": "A.Abusaleh@em.uni-frankfurt.de"},
{"title": "TTLab Paper Accepted at ArabicNLP 2026 StanceEval Shared Task", "content": "We are happy to announce that our paper 'CLASP-Ar: A Cloze-Style Prompting Approach for Arabic-Language Stance Detection' by Bhuvanesh Verma, Ali Abusaleh, and Alexander Mehler has been accepted at the SIGARAB ArabicNLP 2026 StanceEval Shared Task. The paper introduces a cloze-style prompting method for Arabic stance detection.", "date": "2026-09-14", "author": "Bhuvanesh Verma", "email": "verma@em.uni-frankfurt.de"},
{"title": "DUUI Integration of neermatch", "content": "Added a standalone DUUI component for advanced record matching using neermatch as part of CIRCLET. The component integrates deep learning-based entity and record matching into DUUI workflows and includes documentation for installation, model training, and pipeline usage. Support for a ready-to-use baseline model and performance evaluation results has been added and ready to use. Read more: <a href=\"https://py-neer-match.pikappa.eu/\">Neer Match</a> <a href=\"https://github.com/texttechnologylab/duui-uima/tree/main/duui-neer-match\">DUUI Neer Match wrapper</a>", "date": "2026-07-17", "author": "Tim Wolf", "email": "T.Wolf@em.uni-frankfurt.de"},
{"title": "CIRCLET Anonymization Workshop - 8 July 2026 | Please complete our short needs survey by 15 June", "content": "This update invites SPP 2431 project members, affiliated research projects, and researchers working with sensitive, personal, or survey-based data to the CIRCLET Anonymization Workshop hosted by the Text Technology Lab at Goethe University Frankfurt. The hybrid workshop will take place on 8 July 2026 and will address anonymization challenges, GDPR-related foundations, and practical workflows with CIRCLET and DUUI. Interested participants are asked to complete the needs-assessment survey by 15 June 2026: https://anonworkshop.survey.spp.texttechnologylab.org/", "date": "2026-07-08", "author": "Ali Abusaleh", "email": "A.Abusaleh@em.uni-frankfurt.de" },
{"title": "Added a DUUI-Based Component for creating embeddings using self hosted models in Ollama", "content": "This update adds a Dockerized DUUI component for creating sentence embeddings using models hosted in Ollama. This will be later used to create embeddings of documents before adding them to UCE.", "date": "2026-06-03", "author": "Daniel Bundan", "email": "[bundan@em.uni-frankfurt.de](mailto:bundan@em.uni-frankfurt.de)" },
{"title": "DUUI Integration of Climate Classification Models", "content": "Added a new DUUI component for climate-related text classification. The component integrates selected Hugging Face based ClimateBERT models into DUUI and supports climate sentiment analysis, TCFD classification, climate commitment detection, and climate specificity classification. The update includes the new duui-Climate module with Docker support, DUUI integration files, model configuration, documentation, and tests. The supported models include distilroberta-base-climate-sentiment, distilroberta-base-climate-tcfd, distilroberta-base-climate-commitment, and distilroberta-base-climate-specificity.", "date": "2026-05-28", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title": "DUUI Integration of Genre Classification Models", "content": "Added a new DUUI component for genre classification. The component enables document or text segment classification using selected transformer-based genre models. The update includes the new duui-Genre module with Docker support, DUUI Lua and Python integration files, documentation, service scripts, and tests. The integrated models include turkunlp-genre-multi, turkunlp-genre-en, turkunlp-genre-finerweb, ssharoff-genre, and x-genre-classifier, covering both English and multilingual genre classification use cases.", "date": "2026-05-27", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"}, 
{"title": "Six New Sentiment Models for DUUI", "content": "Added six new sentiment models to the DUUI sentiment component. The update extends multilingual and domain-specific sentiment analysis support with finance-focused models for Chinese and French, Twitter sentiment models for Polish, and a climate-related sentiment model for English. The newly added models are finance-sentiment-zh, finance-sentiment-zh-fast, finance-sentiment-fr, twitter-sentiment-pl-base, twitter-sentiment-pl-fast, and distilroberta-base-climate-sentiment.", "date": "2026-05-26", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"}, 
{"title": "Two New DUUI Topic Classification Models", "content": "Added two new topic classification models to the DUUI transformers topic component. The update extends multilingual topic classification capabilities with multilingual-topic-classifier and parlacap-topic-classifier. The changes include updated model documentation, Docker build configuration, requirements, and tests for the DUUI topic pipeline.", "date": "2026-05-26", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"}, 
{"title":"TTLab SPP Website Design Update","content": "Change the Design of the TTLab SPP Website.", "date": "2026-05-28", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},  
{"title":"Per Docker one sentiment model and reproducibility","content": "Every Docker-Container contains only one model, which are offline saved. Added two sentiment models: twitter-xlm-roberta-base-sentiment, citizenlab-twitter-xlm-roberta-base-sentiment-finetunned", "date": "2024-11-06", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title":"Torch Version Update","content": "Updated torch version from 2.2.0 to 2.5.1, because of security issues of the torch python library.", "date": "2025-03-24", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title":"Five new sentiment models","content": "Added 5 new sentiment models: distilbert-base-multilingual-cased-sentiments-student, philschmid-distilbert-base-multilingual-cased-sentiments-student, cardiffnlp-sentiment-en, roberta-based-en, finance-sentiment-de", "date": "2025-03-24", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title":"One new sentiment models","content": "Added one new sentiment models: german-sentiment-bert", "date": "2025-12-05", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title":"Two new topic models","content": "Added two topic models: cardiffnlp/tweet-topic-latest-multi, chkla/parlbert-topic-german", "date": "2024-02-26", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title":"Three new Topic models","content": "Added three topic models: KnutJaegersberg/topic-classification-IPTC-subject-labels, poltextlab/xlm-roberta-large-manifesto-cap, manifestoberta-xlm-roberta", "date": "2024-08-15", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title":"One new topic models","content": "Added one topic models: classla/xlm-roberta-base-multilingual-text-genre-classifier", "date": "2024-08-15", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title":"One new topic models","content": "Added one topic models: tweet-topic-large-multilingual", "date": "2025-11-24", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title":"One new topic models","content": "Added one topic models: topic-organize-web", "date": "2025-04-22", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title":"Removing two Topic models","content": "Removed two topic Models because of security issue and access rights of updatet model: KnutJaegersberg/topic-classification-IPTC-subject-labels, poltextlab/xlm-roberta-large-manifesto-cap", "date": "2024-08-15", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title":"Eight new emotion models","content": "Eight seven emotions Models: finetuned-twitter-xlm-roberta-base-emotion, dreamy-xlm-roberta-emotion, pol_emo_mDeBERTa, xlm-emo-t, emotion-english-distilroberta-base, emotion_text_classifier, cardiffnlp-twitter-roberta-base-emotion, bertweet-base-emotion-analysis", ",date": "2024-03-12", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title":"Eight new emotion models","content": "Eight seven emotions Models: twitter-xlm-roberta, german-emotions, xlm-emo-multi, rubert-cedr-emotion, rubert-tiny2-russian, chinese-emotion-small, chinese-emotion, multi-go-emotions", ",date": "2025-06-10", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title":"17 new emotion models","content": "17 seven emotions Models: zoopa-carer-emotion, distilbert-carer-emotion, corbett-carer-emotion, emo-mobilebert, emopillars-emocontext, universal-joy-multi-small, universal-joy-multi-large, universal-joy-multi-combi, universal-joy-en-huge, universal-joy-en-large, universal-joy-es-large, universal-joy-pt-large, universal-joy-en-small, universal-joy-es-small, universal-joy-pt-small, universal-joy-tl-small, universal-joy-zh-small", ",date": "2025-06-23", "author": "Mevlüt Bagci", "email": "bagci@em.uni-frankfurt.de"},
{"title":"DUUI integration of Multimodal Component","content": "DUUI integration of Multimodal Component. Allows to work with different vLLMs, which support multimodal content. One docker-container can work with multiple vLLMs.", ",ate": "2025-05-06", "author": "Ali Abusaleh", "email": "A.Abusaleh@em.uni-frankfurt.de"},
{"title":"Two new modes and increasing number of supported models (Multimodal Component)","content": "Two new modes and increasing number of supported models(Multimodal Component): vllm/Qwen/Qwen2.5-VL-7B-Instruct, vllm/Qwen/Qwen2.5-VL-3B-Instruct, vllm/Qwen/Qwen2.5-VL-32B-Instruct, vllm/Qwen/Qwen2.5-VL-72B-Instruct, Qwen/Qwen2.5-VL-7B-Instruct, Qwen/Qwen2.5-VL-7B-Instruct-AWQ, Qwen/Qwen2.5-VL-3B-Instruct, Qwen/Qwen2.5-VL-3B-Instruct-AWQ, Qwen/Qwen2.5-VL-32B-Instruct, Qwen/Qwen2.5-VL-32B-Instruct-AWQ, Qwen/Qwen2.5-VL-72B-Instruct, Qwen/Qwen2.5-VL-72B-Instruct-AWQ, Qwen/Qwen2.5-Omni-3B", "date": "2025-05-15", "author": "Ali Abusaleh", "email": "A.Abusaleh@em.uni-frankfurt.de"},
{"title":"New supported models Multimodal Component","content": "New supported models Multimodal Component: Qwen/Qwen2.5-Omni-3B, Qwen/Qwen3-32B, Qwen/Qwen3-14B, Qwen/Qwen3-8B, Qwen/Qwen3-4B, Qwen/Qwen3-0.6B", "date": "2025-05-21", "author": "Ali Abusaleh", "email": "A.Abusaleh@em.uni-frankfurt.de"}
]