|
Title:
|
USER QUERY BEHAVIOR: IMPLICATIONS FOR THE DESIGN OF IMAGE RETRIEVAL SYSTEMS |
|
Author(s):
|
Ya-xi Chen |
|
ISBN:
|
978-972-8924-93-5 |
|
Editors:
|
Pedro Isaías, Bebo White and Miguel Baptista Nunes |
|
Year:
|
2009 |
|
Edition:
|
1 |
|
Keywords:
|
User study, interface design, image retrieval, image tagging. |
|
Type:
|
Full Paper |
|
First Page:
|
527 |
|
Last Page:
|
534 |
|
Language:
|
English |
|
Cover:
|
|
|
Full Contents:
|
click to dowload 
|
|
Paper Abstract:
|
In the last decade, Image Retrieval (IR) has received substantial attention and experienced a rapid growth. While most
research has investigated the actual retrieval algorithms, much less is known about the UI design of IR systems. In this
paper we present the results of an observational user study, in which we observed the users natural behavior in image
searching and tagging. Based on both quantitative and qualitative analysis, we derive implications for the design of IR
systems: Concerning the evaluation of IR systems, besides traditional quantitative evaluation parameters such as recall
and precision, other qualitative evaluation strategies focusing on the users perspective should be integrated to provide a
more user-centered evaluation. IR systems should offer more advanced options such as for sorting, filtering or grouping.
Beyond simple text extraction, an adaptive understanding of the semantics of the query is also required. Because of the
instability, Content-Based Image Retrieval (CBIR) was mainly perceived useful as an inspiration or for discovering
unexpected new results. Tags are a strong alternative to content analysis, but in hybrid IR systems both of them could be
combined to achieve a better performance. There might be a remarkable tagging gap between the indexer and searcher,
and we propose some tagging principles which might help to narrow this gap. |
|
|
|
|
|
|
