東華大學圖書館 |

Computational Fashion Preference Modeling /

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Computational Fashion Preference Modeling // Yunshan Ma.
作者:	Ma, Yunshan,
面頁冊數:	1 electronic resource (138 pages)
附註:	Source: Dissertations Abstracts International, Volume: 84-04, Section: A.
Contained By:	Dissertations Abstracts International84-04A.
標題:	Sparsity. -
電子資源:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29352856
ISBN:	9798352680872

Computational Fashion Preference Modeling /
Ma, Yunshan,

Computational Fashion Preference Modeling /Yunshan Ma. - 1 electronic resource (138 pages)

Source: Dissertations Abstracts International, Volume: 84-04, Section: A.

Human's fashion preference expresses each individual's personalities by adopting or composing a set of preferable fashion elements. Properly understanding and capturing human's fashion preference provide important value to the fashion industry. However, traditional research on fashion preference is mainly carried out by fashion experts, which is highly professional and inefficient as it requires a lot of manually crafted data. With the fast development of the Internet, Data Science, and Artificial Intelligence in recent decades, novel computational solutions are surfacing to revamp this traditional research area, and we term this strand of research as Computational Fashion Preference Modeling. In this thesis, we progressively conduct four studies in computational fashion preference modeling as detailed below.First, we aim to extract human-centered occasion-aware fashion knowledge from social media. Human has accumulated a large amount of general knowledge about what to wear and how to wear, which not only addresses physiological needs but also the requirements in social events and activities. In this study, we propose a novel method to automatically harvest such fashion knowledge from unstructured social media data. Specifically, we propose a contextualized fashion concept learning module to capture the rich context information and a weak label modeling module to effectively utilize the machine-labelled noise data. Even though the extracted fashion knowledge contains rich information, it is still limited in capturing the temporal dynamics and location-aware variance of fashion preference.Second, to address the limitations above, we study the problem of fashion trend forecasting based on social media. Fashion trend forecasting is a crucial task for both academia and industry. Although some efforts have been devoted to tackling this challenging task, they only studied limited fashion elements with highly seasonal or simple patterns, which could hardly reveal the real fashion trends. Towards insightful fashion trend forecasting, this work focuses on investigating fine-grained fashion element trends for specific user groups. In addition, we seek to incorporate rich domain knowledge to enhance forecasting performance. Despite the insights derived from fashion trend forecasting, the granularity of cities and fashion elements is still insufficient when directly applied to real applications such as recommendation.Third, in view of personalized fashion preference modeling, we investigate the problem of personalized sequential fashion recommendation in E-commerce platforms. In order to capture the two types of key patterns, i.e., the user-item interactions and item-item transitions, and address the data sparsity problem, we propose to leverage two types of global graph, i.e., the user-item interaction graph and item-item transition graph. Despite the promising performance in the item-level recommendation, it is incapable to model people's preference over outfits, which is composed of a set of compatible items.Finally, to tackle the problem of personalized outfit recommendation, we are confronted with two types of challenges: 1) more complicated patterns, including user-outfit interaction, user-item interaction, and outfit-item affiliation; and 2) sparse and noisy interactions. To address these challenges, we first propose a light but effective backbone model which can capture the three types of patterns. Second, we introduce contrastive learning as an auxiliary task to deal with the problem of sparse and noisy interactions. In summary, this thesis studies the problem of computational fashion preference modeling from four aspects incrementally, i.e., global and general fashion preference (fashion knowledge extraction), dynamic and locationaware fashion preference (fashion trend forecasting), personalized sequential preference (personalized sequential fashion recommendation), and personalized outfit preference (personalized outfit recommendation). We propose novel approaches to tackling each of the problems, and experimental results on corresponding datasets demonstrate the effectiveness of our methods compared with baselines. Despite of the current achievements, mining and incorporating more comprehensive human fashion knowledge, investigating leading factors from social media, and considering the real industrial needs are promising directions for future work.

English

ISBN: 9798352680872Subjects--Topical Terms:

3680690
Sparsity.

Computational Fashion Preference Modeling /
LDR:05728nmm a22004213i 4500 001 2437634
005 20251124054949.5
006 m o d
007 cr|nu||||||||
008 261011s2022 miu||||||m |||||||eng d
020 $a 9798352680872
035 $a (MiAaPQD)AAI29352856
035 $a (MiAaPQD)USingapore218200
035 $a AAI29352856
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Ma, Yunshan, $e author. $3 3828690
245 1 0 $a Computational Fashion Preference Modeling / $c Yunshan Ma.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2022
300 $a 1 electronic resource (138 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 84-04, Section: A.
500 $a Advisors: Chua, Tat-Seng.
502 $b Ph.D. $c National University of Singapore (Singapore) $d 2022.
520 $a Human's fashion preference expresses each individual's personalities by adopting or composing a set of preferable fashion elements. Properly understanding and capturing human's fashion preference provide important value to the fashion industry. However, traditional research on fashion preference is mainly carried out by fashion experts, which is highly professional and inefficient as it requires a lot of manually crafted data. With the fast development of the Internet, Data Science, and Artificial Intelligence in recent decades, novel computational solutions are surfacing to revamp this traditional research area, and we term this strand of research as Computational Fashion Preference Modeling. In this thesis, we progressively conduct four studies in computational fashion preference modeling as detailed below.First, we aim to extract human-centered occasion-aware fashion knowledge from social media. Human has accumulated a large amount of general knowledge about what to wear and how to wear, which not only addresses physiological needs but also the requirements in social events and activities. In this study, we propose a novel method to automatically harvest such fashion knowledge from unstructured social media data. Specifically, we propose a contextualized fashion concept learning module to capture the rich context information and a weak label modeling module to effectively utilize the machine-labelled noise data. Even though the extracted fashion knowledge contains rich information, it is still limited in capturing the temporal dynamics and location-aware variance of fashion preference.Second, to address the limitations above, we study the problem of fashion trend forecasting based on social media. Fashion trend forecasting is a crucial task for both academia and industry. Although some efforts have been devoted to tackling this challenging task, they only studied limited fashion elements with highly seasonal or simple patterns, which could hardly reveal the real fashion trends. Towards insightful fashion trend forecasting, this work focuses on investigating fine-grained fashion element trends for specific user groups. In addition, we seek to incorporate rich domain knowledge to enhance forecasting performance. Despite the insights derived from fashion trend forecasting, the granularity of cities and fashion elements is still insufficient when directly applied to real applications such as recommendation.Third, in view of personalized fashion preference modeling, we investigate the problem of personalized sequential fashion recommendation in E-commerce platforms. In order to capture the two types of key patterns, i.e., the user-item interactions and item-item transitions, and address the data sparsity problem, we propose to leverage two types of global graph, i.e., the user-item interaction graph and item-item transition graph. Despite the promising performance in the item-level recommendation, it is incapable to model people's preference over outfits, which is composed of a set of compatible items.Finally, to tackle the problem of personalized outfit recommendation, we are confronted with two types of challenges: 1) more complicated patterns, including user-outfit interaction, user-item interaction, and outfit-item affiliation; and 2) sparse and noisy interactions. To address these challenges, we first propose a light but effective backbone model which can capture the three types of patterns. Second, we introduce contrastive learning as an auxiliary task to deal with the problem of sparse and noisy interactions. In summary, this thesis studies the problem of computational fashion preference modeling from four aspects incrementally, i.e., global and general fashion preference (fashion knowledge extraction), dynamic and locationaware fashion preference (fashion trend forecasting), personalized sequential preference (personalized sequential fashion recommendation), and personalized outfit preference (personalized outfit recommendation). We propose novel approaches to tackling each of the problems, and experimental results on corresponding datasets demonstrate the effectiveness of our methods compared with baselines. Despite of the current achievements, mining and incorporating more comprehensive human fashion knowledge, investigating leading factors from social media, and considering the real industrial needs are promising directions for future work.
546 $a English
590 $a School code: 1883
650 4 $a Sparsity. $3 3680690
650 4 $a Forecasting. $3 547120
650 4 $a Multimedia. $2 gtt $3 1000510
650 4 $a User generated content. $3 3562474
650 4 $a Clothing. $3 3682213
650 4 $a Taxonomy. $3 3556303
650 4 $a Data science. $3 3689306
650 4 $a User groups. $3 3696668
650 4 $a Fashion. $3 549143
650 4 $a Physics. $3 516296
650 4 $a Web studies. $3 2122754
690 $a 0800
690 $a 0505
690 $a 0200
690 $a 0796
690 $a 0605
690 $a 0646
710 2 $a National University of Singapore (Singapore). $e degree granting institution. $3 3765631
720 1 $a Chua, Tat-Seng $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 84-04A.
790 $a 1883
791 $a Ph.D.
792 $a 2022
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29352856