Information Retrieval Meeting (IRM 2024)
Information Retrieval Meeting 2024 - the next level: mega databases, artificial intelligence and now?
Mega databases and artificial intelligence are rapidly gaining importance. What does this mean for the future of information retrieval? IQWiG discussed this question at the Information Retrieval Meeting in Cologne on 26 April under the guiding theme: "The next level: mega databases, artificial intelligence and now?" The one-day meeting was the continuation of the format introduced in 2022 by IQWiG's Information Management Department.
More than 100 national and international experts, mainly from HTA agencies and academia, came to Cologne to discuss artificial intelligence (AI), mega databases, new technical tools, and the foreseeable consequences for information retrieval. In plenary sessions, workshops and personal networking, participants had the opportunity to learn about modern digital options and thus gain important new insights for their daily work.
Aaron Tay
Aaron Tay from Singapore Management University kicked things off with his keynote speech on the trends that could drastically change information retrieval in the next decade. He presented mega databases such as Semantic Scholar, Lens.org and OpenAlex as well as GPT-based AI tools. He noted that especially the development of the semantic search, where systems understand the meaning of sentences, is something to keep an eye on.
Harrisen Scells:
In his workshop, Harrisen Scells from the University of Leipzig then explained what options Application Programming Interfaces (APIs) already offer. He noted that APIs are a powerful tool that can be used to support a wide range of information retrieval tasks. The basic components of an API query are easy to understand and follow. In particular, the interfaces to thesauri (MeSH) and bibliographic databases as well as search platforms (PubMed, Ovid) could speed up and simplify the development of search strategies.
Ian Shemilt:
In his workshop, Ian Shemilt from the EPPI Centre at University College London presented the latest advances in using OpenAlex to efficiently and automatically update systematic reviews. He demonstrated the integration of OpenAlex using the EPPI reviewer. As a practical example, he used the IQWiG report V22-06A "Evidence search for the S3 guideline on the diagnosis and treatment of gout". The references finally included after full-text screening and the excluded references formed the basis of his analysis. The integration of OpenAlex makes it possible to access new references and preprints on a reference or topic basis. However, he pointed out that it is not yet clear whether it will be possible to update Cochrane reviews in this way in the future.
Rabea Müller:
In the afternoon, Rabea Müller from the Central Library of Medicine in Cologne presented a workshop on tools for libraries and information specialists (Library Carpentries), focusing on three programmes: the dynamic and versatile programming language Python, the version management software Git, and the command line tool Unix Shell for Unix operating systems. The latter in particular allows users to perform complex and efficient tasks with little effort. She emphasized that repetitive tasks can thus be automated and smaller tasks can be combined into larger, more efficient workflows. Within the Library Carpentries, workshops are offered on topics such as APIs.
Amy Finnegan:
In her workshop, Amy Finnegan from the British HTA agency NICE presented a pilot project for the automated exclusion of animal studies (classifier based on pattern matching) that could not be optimally excluded in advance by animal filters (limitations) in searches for scientific literature. She emphasized that she did not focus on database classifications, but worked with her own high-quality training data. This project was embedded in an "IS upskilling" project, in which 6 of the 35 information specialists (IS) at NICE took part in a 12- month programming course (programming language R). 5 of the course participants had no prior programming knowledge and were given one working day a week for this project work. In total, NICE completed 5 projects under this umbrella project (including data reformatting tools, automatic DOI retrieval, etc.). Initially, smaller projects were undertaken to gain a basic understanding of the use and implementation of R in order to be able to develop classifiers for non-RCTs in the future, for example.
Fishbowl discussion:
The Information Retrieval Meeting concluded with a joint fishbowl discussion on questions about the future of information retrieval: Will the new tools possibly make information specialists obsolete or just change the job profile? And if so, how? A very interesting discussion ensued, with the participants agreeing that the job profile is likely to change, but will probably become even more important in the future.