Important Dates


NTCIR-14 Lifelog Access and Retrieval (Lifelog) Task



=== OVERVIEW ===

Enhancing personal health and wellness is one of the global challenges being faced by society. Naturally data (especially personal data) will form a major component of any advancements in the area. At NTCIR-14, the Lifelog task is a starting point for addressing such challenges and provides a forum for researchers to compare their approaches with state-of-the-art approaches from their peers in and open and competitive environment NTCIR is an ongoing series of workshops focusing on a list of different information retrieval (IR) research areas and this is the 14th instantiation of the series, with numerous tasks, including the Lifelog task.

The Lifelog task introduces a new dataset (45 days in total) of rich lifelog data from a number of individuals, including all-day wearable camera data, manually captured foodlogs, rich biometric data, continuous blood glucose monitoring data, location data and details of physical activities. This data is enriched with state-of-the-art visual annotations for the wearable camera data. This is the richest dataset of human activity data ever released. There are three sub-tasks, any (or all) of which can be undertaken by participating teams, as highlighted below. All participating teams are expected to write a paper on their work (accepted papers are published by Springer LNCS) and attend the three day NTCIR workshop in Tokyo in June 2019, or a co-timed satellite event in Sheffield, UK.

=== TASKS ===

NTCIR-14 Lifelog includes three subtasks, all of which can be participated in independently.

LADT (Lifelog Activity Detection subTask)
The aim of this subtask is to new approaches to the annotation of multimodal lifelog data in terms of activities of daily living. An ontology will be generated of important lifelog activities of daily living and the task will require the development of automated approaches to annotating these concepts. Both image content as well as provided metadata and external evidence sources can be used to generate the annotations. Once again, there will be a focus on many health-related activities.

LSAT (Lifelog Semantic Access subTask)
In this subtask, the participants have to retrieve a number of specific moments in a lifelogger's life. This can be achieved by both automatic and interactive retrieval systems. We define moments as semantic events, or activities that happened throughout the day and there will be 16 moment queries released with the test collection. Relevance judgements will be released to participants after submission of runs. The task can best be compared to a conventional known-item search task.

LIT (Lifelog Insights subTask)
Based on the Quantified Self movement, the LIT subtask asks participants to generate new types of visualisations and insights about the life of lifeloggers based on the source data. The focus is on health & wellness, and we propose that participants explore the data to generate insights based on diet, exercise levels, and across-participant comparisons. This sub-task is not evaluated in the traditional sense, but the insights (and production methods) are presented at the workshop and described in the proceedings paper.

Please visit NTCIR lifelog website for more information about the task:


Cathal Gurrin (Dublin City University, Ireland)
Hideo Joho (University of Tsukuba, Japan)
Frank Hopfgartner (University of Glasgow, UK)
Liting Zhou (Dublin City University, Ireland)
Duc-Tien Dang-Nguyen (Dublin City University, Ireland)

=== CONTACT ===

ntcir-lifelog at computing dot dcu dot ie