Understanding and exploiting social media data to support decision making in fashion and apparel supply chains
|Authors:||Beheshti-Kashi, Samaneh||Supervisor:||Thoben, Klaus-Dieter||1. Expert:||Thoben, Klaus-Dieter||Experts:||Dr. Lawo, Michael||Abstract:||
Fashion and apparel supply chains are highly dynamic, complex and challenged by high uncertainties. Stakeholders are typically confronted with incomplete information at the time when they do require it. For complementing this lack of information, additional sources are explored involving historical sales data. However, for fashion items historical sales data are often sparse. Ever since the emergence of social media, its relevance for fashion and apparel supply chains keeps rising. The exploitation of social media should not only be considered as a purely technical task, as social media comes with a range of characteristics. Being classified as a big data source, the 5 V's apply also to social media. In addition to volume, velocity, variety of data, the veracity feature is fundamental in terms of using the extracted information for supporting decision processes. Nevertheless, most existing approaches which use social media data for fashion and apparel problems do not consider neither these characteristics nor fashion and apparel supply chain characteristics. In addition, stakeholders' needs and requirements are neglected when exploiting social media data. This said, the present research claims the necessity of a methodology for exploiting social media data for fashion and apparel supply chain decisions considering the stakeholders' perspective. In this manner, this thesis poses the question if and how social media can be used as an addition information source to support fashion and apparel supply chain decisions.
For targeting this research question, as an overall framework a ''Design Science Research'' approach is selected. Following the ''Design Science Research Methodology Process Model'', this thesis first designs and develops an artifact, designed as a process model enabling an understanding of social media data and a systematical exploitation of textual social media data to support fashion and apparel SCs decisions, secondly, shows its use in a demonstration case and finally evaluates its utility by the judgment of experts working in the field of fashion and apparel supply chains. For the development of the process model, characteristics of fashion and apparel supply chains and social media, supply chain stakeholders, as well as text mining and process model design features are considered. Accordingly, the process model consists of four layers: the Process Layer, the Information Source Layer, the Social Media Layer and the Text Mining Layer. The realization of the process model is conducted in Business Process Model and Notation 2.0.
Following the ''Design Science Research Methodology Process Model'', the utility of the artifact designed is shown in a demonstration by implementing a case study around the product feature colour. The process model is applied on blog data. In order to show the feasibility of social media data as an additional information source, the existence of an economically advantageous time offset between sales and blog data is examined. The findings show that it is possible to identify an advantageous economic value for two colour groups, even over different supply chain stages. Having demonstrated the utility of the process model by the case study, the evaluation is performed based on its results. This involves comparing the objectives of the artifact with the oberserved results generated from its use. An ex post naturalistic approach is applied and manufacturers and retailers from fashion and apparel supply chains are surveyed. The evaluation has demonstrated that supply chain stakeholders see an added value from the extracted information in particular when available two or four months in advance of the selling season.
|Keywords:||Fashion and Apparel Supply Chains; Social Media data exploitation; Believability Assessment; Process Model||Issue Date:||30-Nov-2020||Type:||Dissertation||DOI:||10.26092/elib/1105||URN:||urn:nbn:de:gbv:46-elib53205||Institution:||Universität Bremen||Faculty:||Fachbereich 03: Mathematik/Informatik (FB 03)|
|Appears in Collections:||Dissertationen|
checked on Feb 2, 2023
checked on Feb 2, 2023
Items in Media are protected by copyright, with all rights reserved, unless otherwise indicated.