The browser you are using is not supported by this website. All versions of Internet Explorer are no longer supported, either by us or Microsoft (read more here: https://www.microsoft.com/en-us/microsoft-365/windows/end-of-ie-support).

Please use a modern browser to fully experience our website, such as the newest versions of Edge, Chrome, Firefox or Safari etc.

Describe AI research at Lund University with AI (ALI 2-05)

DRAFT • DRAFT •DRAFT •DRAFT •DRAFT •DRAFT •DRAFT •DRAFT •DRAFT

Here you may sign up to contribute with your expertise or opt out.

On this web page the project AI research at Lund University with AI

  • inform about the project, 
  • explains how researchers at Lund university should do if they want to opt-out,
  • explains how AI oriented researchers at Lund University can contribute to the project by validating quality of the  output from the AI system within their area of expertise.

Project title in Swedish: Beskriva AI Forskning vid Lunds universitet med AI

Period: October 2024 to June 2025

Project Description in Swedish: PDF

Integrity protection plan in Swedish: PDF

Project team
Peter Berk, Humanities Laboratory, Lund university
Jonas Wisbrant, AI Lund

Reference group:
TBD

Background and purpose: There is a lot of research going on at Lund University. Especially in the field of Artificial Intelligence (AI), the number of articles and contributions from individuals at Lund University has increased. We want to investigate a method to capture and make research visible by analysing data from LUCRIS. As development has progressed, language models are starting to reach the point where they are useful for semantic indexing, processing, and searching content. We want to investigate what these models and related techniques can do to e.g. suggest relevant researchers and research at Lund University.

In summary, the project has three objectives: 

  • Investigate whether language models can be used to map AI research 
  • Test and disseminate a work process for legally secure use of language models on personal data.
  • Disseminate skills and knowledge within universities and the outside world about how to validate research. 

The project: The main idea of the project is to create models, user interfaces and other technical solutions that analyze summaries of articles, participating authors and institutions and similar data linked to these to investigate what networks and trends exist. Possible applications for evaluation may include the following use cases:

  1. to find through free-form questions and summaries which people at Lund University best match or have some relevance to a question asked. 
  2. to be able to perform semantic searches based on a short description and produce a list of relevant articles, institutions and individuals that are linked to the research question 
  3. to be able to evaluate existing and suggest new keywords that describe an article,
  4. to propose a summary of selected research, etc. based on, for example, authors, organisation or research project.

The project is carried out in dialogue with the system owners of LUCRIS and LU's data protection officer and NLP researchers at LU. 

Compute capacity and data storage are primarily requisitioned from LU-controlled environments. Project phases: 

  1. Integrity plan, Data requisition from LUCRIS, Setting up development environment. 
  2. Communication with researchers affected by personal data, and recruitment of reserchers for expert validation of results.
  3. Development of prototype systems. 
  4. Validation together with experts (researchers at LU) in surveys or workshop format. 
  5. Project evaluation and communication of experiences. Networking and dissemination of experience: All researchers are addressed to provide the opportunity: to opt-out in their own research. to participate in the validation of the system by assessing the system's ability to properly illustrate their research.

Processes and documents for legally secure research with a language model are published and disseminated internally and publicly via AI Lund and other forums such as lunch seminars and websites.

Contribute with your expertise

The project will start out with use case four above. For that we need help from researchers from al  faculties that are active in the active in the AI domain. If you sign up to contribute we expect you to spend a one to three hours on:

  • evaluate the AI systems description of your research in the AI area
  • compare the AI systems description of AI research for specific researchers or on organisation level to corresponding  existing text written by humans.

If you sign up to contribute you will in the web form be asked to describe your research in the AI area with 150 to 500 words. The text will be used a for test and and benchmarks when prototyping tha AI system and later on for comparison to  other texts.

Opt-out your publications

description of consequenses

[webform here]

Last updated: 8 Oct 2024