The Research Studio Data Science develops a new information access system to properly address what is missing in a professional user’s knowledge base and consequently present him/her with suitable and valuable information.
Dealing with the information needed by scientifically and technically educated users is often a challenge for existing information retrieval systems. One of those challenges is due to the fact that general purpose information retrieval system do not take into account the user’s knowledge in the retrieval models, therefore potentially offering either obvious or too specialized resources.
DoSSIER‘s goal is to address this issue by advancing the state-of-the-art in our understanding of computer-supported human information search workflows. The result of this project will be a new generation of information access systems, which will accelerate innovation cycles in EU academia and industry. Learn more on Dossier-Project.eu.
An excellent and highly synergistic team of world leading Information Retrieval (IR) experts from 5 EU countries, together with 3 academic partners (universities in the US, Japan and Australia) and 11 industrial partners are exploring fundamental insights into how users understand, formulate and access information in professional environments.
Project 13 of DoSSIER (also known under the code KD; i.e. knowledge delta) aims at modelling the knowledge of different users and the available documents and thus solving the problem of scientific information retrieval by finding the most relevant documents for each user. The difference between what a user knows and the knowledge he is missing, which exists in the documents on the web, is defined as knowledge delta. The knowledge delta for each user and domain is specific, so KD has two main goals:
1) Definition and model of textual similarity as knowledge difference
2) To clarify the relationship between information needs and linguistic representations
Yasin Ghafourian, a new Researcher at the Research Studio DSc, was selected to carry out the project KD under the supervision of Dr. Petr Knoth, director of the Research Studio DSc, and Prof. Allan Hanbury from the Vienna University of Technology:
“Within the project KD we will see that a new formal definition of a user’s knowledge, based on the merging of statistical, latent, and explicit semantics that also incorporated user’s long term and short term information interests is developed. Subsequently, having set up a model for user knowledge we will then initiate a novel information retrieval model that is based not solely on the similarity between the linguistic expression of an information need and the content of indexed documents, but also, and primarily, on the difference between the knowledge model of the user and that of the indexed collection and thereby making sure that each professional user can be presented only with information that best suits her according to her current knowledge of the field and interests.” Yasin Ghafourian about his new project.
More about Yasin Ghafourian
Yasin Ghafourian received his Bachelor’s degree in Computer Software Engineering from Kharazmi University in Tehran in 2016 and his Master’s degree from the University of Tehran in Iran in 2019. He also has one year of experience as a data/business analyst for an Iranian online travel agency. In terms of his research in information retrieval, he has worked on improving question retrieval algorithms in the context of relevance models in community-based question answering systems, gaining experience in working with relevance models, expansion algorithms, word-embedding based models and artificial deep neural networks.