vLLM automatic Sentimant-Topic-Emotion classification - part of TTLabTweetCorpus
Year: 2026
Type: software
Abstract: This work is part of TTLabTweetCorpus This tool is part of TTLabTweetCorpus, please refer to the full tool usage in the following link https://github.com/texttechnologylab/TTLabTweetCorpus. Java tool for running multimodal (image/video) LLM annotation over tweet datasets using DUUI (Docker Unified UIMA Interface) and the duui-mm component. For each tweet, the attached image or video is sent to a vision-language model (e.g. Qwen2.5-VL, Qwen3-Omni) which returns a structured JSON annotation covering: Emotion classification (anger, sadness, apprehension, confusion, happiness) Sentiment analysis (positive, neutral, negative) Topic classification (Dewey Decimal Classification, level 1) Each label is returned with a confidence distribution and a short evidence-based explanation. Note: When used, the citation should match the citation mentioned in https://github.com/texttechnologylab/TTLabTweetCorpus/blob/main/CITATION.bib as follows: @inproceedings{bagci:et:al:2026, title = {Extending a Parliamentary Corpus with {MP}s{\textquoteright} Tweets: Automatic Annotation and Evaluation Using {TTLABTWEETCORPUS}}, author = {Mevlüt Bagci and Ali Abusaleh and Daniel Baumartz and Alexander Mehler and Giuseppe Abrami and Maxim Konca}, booktitle = {KONVENS 2026 - Context matters: NLP beyond Text}, year = {2026}, address = {Hamburg, Germany}, keywords = {Corpora, Tweets, Political Data, Text, Media classification}, note = {Accepted}}
Author: Abusaleh, Ali
Institution: Goethe University Frankfurt
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7223779080