EmergentIR: Improving Informational Web Search for Emerging Topics

• Applicants: Professor Dr. Stefan Dietze; Dr. Ran Yu
• Subject Area: Data Management, Data-Intensive Systems, Computer Science Methods in Business Informatics
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 548295069

Informational Web search is among the most frequent online activities, where a user aims at acquiring knowledge with respect to a particular topic. As demonstrated during the COVID-19 pandemic, misinformation on the Web in particular in the context of emerging topics had substantial impact on public opinion, knowledge and actual behaviour. Contemporary search engines, however, are optimised towards retrieval of relevant documents for a given information need but are not well-suited to efficiently aid learning and knowledge acquisition on emerging topics. Limitations of search engines in this context arise from their reliance on data, such as relevant Web documents to be retrieved and ranked, historical query and click-through data, link graphs serving as indicators of authority and popularity, and background knowledge from trusted sources such as corresponding Wikipedia entries or the Google knowledge graph. For emerging topics, data in all categories tends to be both sparse and dynamic, making them less reliable, more diverse with respect to quality and trustworthiness and prone to deliberate or accidental manipulation, misinformation and bias. Furthermore, during informational search sessions on emerging topics, merely ranking resources is not the most efficient way of aiding knowledge gain, as users typically require condensed summaries, structured background knowledge or facts in order to categorize, contextualise and comprehend new information. Due to the aforementioned reasons, aiding users in efficiently digesting information is both particularly important and particularly challenging in the context of emerging topics. Our objective is to improve user knowledge gain in informational Web search sessions on emerging topics through evaluating and improving state of the art of methods aimed at i) detecting search session intents associated with emerging topics as opposed to non-emerging topics, ii) robust ranking of relevant Web resources from a potentially sparse and dynamically evolving candidate pool in the context of emerging topics and, iii) summarising information to help users gaining knowledge more efficiently on the searched emerging topics. We will follow a user-centric approach, that focuses specifically on evaluating the ability of established IR-algorithms to facilitate knowledge gain of users on emerging topics. We will build on prior works at the intersection of learning theory, information retrieval, summary and fusion of knowledge and information and the prediction of knowledge gain during Web search.

• DFG Programme: Research Grants