دانلود رایگان مقاله انگلیسی رویکرد تشخیص خصوصیات موثر محصول مبتنی بر وب کاوی در محیط جمع سپاری - تیلور و فرانسیس 2018

ABSTRACT

Affective product design, which aims to satisfy customer feelings as an aspect of product quality, has attracted more and more research attention. However, conventional product design relies on surveys and user experiments to collect user evaluations, which leads to the issues that (i) consumers can only express their feelings on design attributes specified by assigners; (ii) abundant online consumer resources are neglected; and (iii) a lack of further prioritisation and re-construction of affective design properties. This study aims to develop a product affective properties identification approach. Crowdsourcing platforms have the advantage of obtaining large numbers of free consumer comments and have been utilised as data sources. Web mining and text mining are deployed to capture the crowdsourced product review resources. Then product design knowledge hierarchy is established to find design properties, while sentiment analysis was undertaken to identify affections. With the help of domain ontology to connect design properties and corresponding affections, product affective properties can be identified. Furthermore, the identified affective properties are prioritised, so as to assist in design improvement and support decision making. To illustrate the proposed approach, a pilot study on iPhone 7 was conducted, in which influential affective properties have been identified and ranked.

5. Discussion and conclusions

This work aims to develop a product affective property identification approach. For this end, a web- and text-mining process is deployed to make use of online product review resources, capture useful consumer responses and perform textual analysis. Afterwards, design knowledge hierarchy is constructed to support the identification of design-related tokens, and sentiment dictionaries are utilised to identify affect-related word tokens. With the help of domain ontology and electronic lexical database to provide semantic relations and lexical reference, the associations between design tokens and affective tokens can be examined. The design properties which are related with affective tokens are regarded as affective design properties. The design importance and affective intensity of the affective properties are estimated, and overall priority of these properties can be accordingly achieved. Through a pilot study, it has been demonstrated that the proposed approach is capable of capturing more affective design properties, and a clear and practical reference in terms of the priorities of different design properties can be achieved so as to facilitate decision making and product improvement.

However, there are still some limitations of this work. For example, the identification of affection is based on sentiment analysis, which actually relies on the recognition and measurement of affective words (which have been defined and tagged in existing lexical databases or sentiment dictionaries). If consumers’ statements do not include obvious affective words, the feeling may not be detected. Moreover, since the focus of this work is the identification of affective design properties, consumers’ emotions are not investigated in very specific types, such as happiness, sadness, confidence and confusion. In future research, consumers’ emotions will be further studied. In addition, the redesign strategy based on the product affective features will be explored to facilitate design practice in order to fulfil users’ affective expectation and achieve successful product design.