[visionlist] Call-for-Participation: 1st Recommending Task @ ImageCLEF 2023 (Cultural Heritage Content-based Recommendation Task)

[Bogdan Ionescu]

[Apologies for multiple postings]

ImageCLEFrecommending (1st edition)
Registration: https://www.imageclef.org/2023/recommending
Run submission: May 10, 2023
Working notes submission: June 5, 2023
CLEF 2023 conference: September 18-21, Thessaloniki, Greece


*** CALL FOR PARTICIPATION ***
In recent years cultural heritage organisations have made considerable
efforts to digitise their collections, and this trend is expected to
continue due to organisational goals and national cultural policies.
Thus media archives have not only exponentially increased in size, but
now hold contents in various modalities, e.g., video, image, text.
Even when structured metadata is available it is still difficult to
discover the contents of media archives and allow users to navigate
multiperspectivity in media collections. Content-based recommendation
systems can help but there is limited understanding how well these
perform and how relevant they are for the end-users. Moreover, the
system used so far have not addressed the new user requirements of
more transparency and explainability of the algorithms used.

The task targets a key infrastructure for researchers and heritage
professionals: Europeana. With over 53 million records, the single
search bar that served as the main access point was identified as a
bottleneck by many users. Thus, the strategy has gradually shifted
towards exploration of the available collections based on themes. Now
users can explore over 60 curated digital exhibitions, countless
galleries and blog posts. While there is a system in place to
recommend individual items given a query item, the recommendations for
editorials are done at the moment only manually. For instance when a
new blog is created, the author would manually provide a list of
related galleries, blogs or exhibitions that have been already
published.


*** TASK ***
The task requires participants to devise recommendation methods and
systems, apply them in the supplied data set gathered from Europeana
and provide a series of recommendations in two scenarios: (i) given a
list of items provide a list of recommended items; (ii) given an
editorial (Europeana blog or gallery) provide a list of recommended
editorials.


*** DATA SET ***
For the task a new dataset based on Europeana items and editorials
will be provided to the participants. The individual items in the
dataset will include a wealth of metadata based on the Europeana Data
Model (EDM) schema. Editorials will be either Galleries (containing a
title, optional description and list of items which make it up), or
blog posts (containing a title, text in English and a number of
items). It should be noted that although all data items follow EDM the
quality of the metadata is not perfect, with some data fields being
potentially somewhat ambiguous, or at least used sometimes in a
creative way by the original data providers (especially with some
overlap sometimes in what ends up in "format", "medium", "type" and
"subject".


*** METRICS ***
Performance will be evaluated on the basis of the recommendations that
are provided computing Mean Average Precision at X (Map at X) compared to
the ground truth. Moreover, the systems competing in this task that
can provide an explanation for the results provided will be preferred.


*** IMPORTANT DATES ***
- Run submission: May 10, 2023
- Working notes submission: June 5, 2023
- CLEF 2023 conference: September 18-21, Thessaloniki, Greece
(https://clef2023.clef-initiative.eu/)


*** OVERALL COORDINATION ***
Alexandru Stan, IN2 Digital Innovations, Germany
George Ioannidis, IN2 Digital Innovations, Germany
Bogdan Ionescu, Politehnica University of Bucharest, Romania
Hugo Manguinhas, Europeana Foundation, Netherlands


*** ACKNOWLEDGEMENT ***
The task is supported under the H2020 AI4Media “A European
Excellence Centre for Media, Society and Democracy† project,
contract #951911 https://www.ai4media.eu/.


On behalf of the Organizers,

Bogdan Ionescu
https://www.AIMultimediaLab.ro/