@hiemstra I settled up my ML learning to rank architecture for the part 2&3 but unfortunately I can't find enough relevant data for training (nothing on TREC related to searching in PubMed and the index we have is different from the one online the PubMed website so I can't use their one :( ). Do you have other relevance judgment than the 50 we already have ?
Or shall I try something else ?

@tdouzon We don't have more training data for machine learning, unfortunately. You might try the following. In absence of training data, Mostafa Deghani and colleagues used "weak supervision" with BM25 as "signal", see: Mostafa Dehghani et al. "Neural Ranking Models with Weak Supervision", In Proceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR'17), 2017, pp 65-74.

Sign in to participate in the conversation
Mastodon U. Twente

A social network for the University of Twente community. Anyone with an or @* email address can create an account here and participate in the global fediverse with millions of other users. This means that students, staff and alumni can create an account here. Content does not reflect the opinions or policies of the University of Twente.
We support \( \LaTeX \) formulas: Use \( and \) for inline LaTeX formulas, and \[ and \] for display mode.