Stay tuned for LongEval 2025 announcements !
- -Agenda for September 12 (all times EST)
+ +Agenda for September XX (to change this section) (all times EST)
| 12:25 - 12:45 | -Team Galápagos Tortoise at LongEval 2024: Neural Re-Ranking and Rank Fusion for Temporal Stability + | Team Galápagos Tortoise at LongEval 2025: Neural Re-Ranking and Rank Fusion for Temporal Stability
Marlene Gründel, Malte Weber, Johannes Franke and Jan Heinrich Merker |
@@ -91,16 +90,23 @@
LongEval Lab
-In this page we present CLEF 2024 shared task evaluating the temporal persistence of information retrieval (IR) systems and text classifiers. The task is motivated by recent research showing that the performance of these models drops as the test data becomes more distant in time from the training data. LongEval differs from traditional IR and classification shared task with special considerations on evaluating models that mitigate performance drop over time. We envisage that this task will bring more attention from the NLP community to the problem of temporal generalisability of models, what enables or prevents it, potential solutions and limitations.
-The CLEF 2024 LongEval Lab encourages participants to develop temporal information retrieval systems and longitudinal text classifiers that survive through dynamic temporal text changes, introducing time as a new dimension for ranking models performance.
+LongEval CLEF 2025
+LongEval at CLEF 2025 is the third edition of the shared task focusing on the temporal persistence of information retrieval (IR) models. The goal is to evaluate how retrieval models perform as the test data evolves over time, emphasizing the challenge of maintaining retrieval quality despite changes in user queries and document relevance.
+ +This year, LongEval introduces two retrieval tasks:
+ +Task 1: LongEval-WebRetrieval
+A continuation of the Web retrieval challenge, using evolving datasets from the French search engine Qwant. Participants develop IR models that maintain performance across different temporal lags.
+ +Task 2: LongEval-SciRetrieval
+New in 2025, this task evaluates retrieval performance in scientific search using data from the CORE repository, the largest collection of Open Access scholarly documents. Systems are tested on queries and relevance feedback from real users.
Previous year
- Check our 2023 website to find information about previous years. +Check our 2024 website and 2023 website to find information about previous years.
-Contact
-For Task 1. LongEval-Retrieval: longeval-ir-task@univ-grenoble-alpes.fr
-For Task 2. LongEval-Classification: Rabab Alkhalifa
+Contact XX to be same of change?
+For Task 1. LongEval-WebRetrieval: longeval-ir-task@univ-grenoble-alpes.fr
+For Task 2. LongEval-SciRetrieval: Rabab Alkhalifa
Join our slack channel for any question.