In the field of information retrieval (IR), harnessing knowledge from external knowledge bases, such
as Knowledge Graphs and Wikipedia, has emerged as a promising avenue for improving the effectiveness
and the interpretability of IR systems. These external knowledge bases offer valuable information
that empowers IR models to make more accurate predictions. The infusion of external knowledge bases
into IR models can provide enhanced ranking results and greater interpretability, offering
substantial advancements in the field.
The First Knowledge-Enhanced Information Retrieval workshop (KEIR @ ECIR 2024) will serve as a
platform to bring together researchers from academia and industry to explore and discuss various
aspects of knowledge-enhanced information retrieval systems, such as models, techniques, data
collection and evaluation. The workshop aims to not only deliberate upon the advantages and hurdles
intrinsic to the development of knowledge-enhanced PLMs, IR models and RecSys models but also to
facilitate in-depth discussions concerning the same.
This workshop holds a steadfast commitment to fostering collaboration among researchers engaged in the realm of knowledge integration for IR, RecSys and NLP. We invite submissions regarding different aspects of knowledge-enhanced information retrieval systems. Relevant topics include, but are not limited to:
We invite authors to submit papers written in English. Submissions may range in length from a minimum of 6 pages to a maximum of 12 pages; however, references and supplementary materials may exceed this page count without limitation. In order to facilitate a double-blind review process, authors must ensure that submissions are fully anonymized. Please note that we do not impose a specific anonymity period prior to submission. The papers (.pdf format) should be submitted using the EasyChair submission system at https://easychair.org/conferences/?conf=keirecir2024. Authors should consult Springer’s authors’ guidelines and use their proceedings templates to prepare the submission. The Microsoft Word and LaTeX versions of the template can be found at https://www.springer.com/gp/computer-science/lncs/conference-proceedings-guidelines. Submissions to KEIR@ECIR2024 will be peer-reviewed on the basis of technical quality, relevance to workshop topics, originality, significance, clarity, etc. We accept submissions of the following types:
The proceedings of the KEIR workshop will be non-archival in nature. Extended versions of selected papers will be recommended for special issues of several SCI-indexed international journals. However, authors retain the right to submit their work to other peer-reviewed venues for further dissemination.
Abstract
Conversational Recommender Systems (CRS) have revolutionized how we interact with the recommender by discerning user preferences through iterative dialogues. However, the effectiveness of these systems is often hindered by a need for more contextual understanding for precise preference modeling. Integrating various external knowledge sources offers promising solutions to enrich dialogue context, predict user preferences, and generate informative responses. This presentation will delve into recent advancements in the realm of knowledge-enhanced CRS, spotlighting studies that leverage knowledge to refine recommendations and enrich user interactions. I will share insights from our recent research that bolsters the recommendation quality and the informativeness of conversational responses. Additionally, I will illuminate emerging research trajectories within this domain, emphasizing the synergy between CRS and Large Language Models.
Bio
Dr. Zhaochun Ren is an Associate Professor at Leiden University, the Netherlands. He is interested in information retrieval and natural language processing, with an emphasis on conversational artificial intelligence, recommender systems, and social media analysis. He aims to develop intelligent agents that can address complex user requests and solve core challenges in NLP and IR towards that goal.
Abstract
At Bloomberg, we build search applications for financial professionals as a part of our main product, the Bloomberg Terminal. In this talk, we will provide an illustration of a typical search workflow at Bloomberg, and discuss 1) how structured and encoded knowledge could be utilized end-to-end in a search system, and 2) challenges in applying them effectively at each step. Early in the pipeline, knowledge can be used to enrich and organize documents in the searchable collection. Thus, we start by discussing the document understanding step. Then, we consider the challenges of interpreting queries correctly and applying the appropriate retrieval configuration needed to satisfy the user’s information needs. We then continue to discuss the utilization of such enriched knowledge for document retrieval and reranking. Lastly, since user experience (UX) plays an important role in the search process, we will also discuss open questions around designing the user experience that is guided by knowledge.
Bio
Dr. Ridho Reinanda is an AI Research Scientist who is currently leading the Knowledge Graph team at Bloomberg. He obtained his Ph.D. in Information Retrieval at the University of Amsterdam, where he focused on leveraging knowledge graphs for information retrieval tasks and applying IR techniques for knowledge graph maintenance.
Activity type | Time | Activity |
---|---|---|
Opening remarks | 9:00 - 9:15 | |
Keynote I | 9:15 - 10:05 | Keynote I by Dr. Ridho Reinanda (45 min Talk + 5 min Q&A) |
Invited talks | 10:05 - 10:30 | Invited talk by (1) Shubham Chatterjee (20 min Talk + 5 min Q&A) |
Coffee break | 10:30 - 11:00 | |
Paper presentations | 11:00 - 12:30 | Accepted paper presentations (1-4) (each 15 min Talk + 5 min Q&A) |
Lunch | 12:30 - 13:30 | |
Keynote II | 13:30 - 14:20 | Keynote II Assoc. Prof. Zhaochun Ren (45 min Talk + 5 min Q&A) |
Invited talks | 14:20 - 15:10 | Invited talks by (2) Paul Lerner and (3) Zeyuan Meng (each 20 min Talk + 5 min Q&A) |
Coffee break | 15:10 - 15:30 | |
Invited talks | 15:30 - 16:20 | Invited talks by (4) Sara Abdollahi and (5) Erlend Frayling (each 20 min Talk + 5 min Q&A) |
Closing remarks | 16:20 - 16:30 |