Wikimedia+Libraries International Convention 2025/Programme/nlp-information-literacy
Natural Language Processing (NLP) for empowering information literacy of library users with regard to disinformation- EVA SEIDLMAYER
Time: 15:30-15:45
Room: Alfonso Reyes Auditorium
Abstract:
The proliferation of deliberately spread disinformation presents a substantial threat to democratic societies. Beyond political assertions, scientific domains, particularly health and life sciences, are significantly impacted by this phenomenon.
The contribution introduces a dual-pronged approach to countering disinformative content within life sciences. The first component involves classification based on semantic features through the training of a language model. The second component focuses on Named Entity Recognition (NER), employing established techniques to identify relevant concepts in questionable texts. These concepts can then be linked to Wikidata items, allowing the display of contextual Wikidata information alongside the original text.
Both sub-components aim to provide users with supplementary information, rather than engaging in fact-checking or definitive assessments, which risk allegations of censorship. The goal is to empower users, enabling them to make informed decisions independently.