以情緒分析探討使用者評論對商品未來設計之影響

Finding Product Future Design Based On Emotional Analysis On Product Review

蘇達立(TA-LI SU)

指導教授 ： 謝俊霖

國立臺灣大學/管理學院/資訊管理學研究所/碩士(2021年)

https://doi.org/10.6342/NTU202102165

摘要

本論文的研究動機旨在透過評論中隱藏的情緒來找出這些情緒對於未來商品設計之影響，本論文以使用者評論為資料，以情緒分析的方式找出未來影響商品設計之關鍵情緒，本論文主要的商品為智慧型手機市場，使用者評論的平台為amazon、gsmarena,使用之情緒分類模型為BERT，再經由情緒分析後發現，憤怒的情緒不管是在市佔率以及未來產品功能皆相較於其他情緒有更大的影響。本次的研究相較於之前相關領域的研究不同的是，本論文以情緒分析作為判定情緒的方法，相較以往以情感分析的方式更為的細緻，另外在評論重要程度的方面，本論文參考了先前的研究使用了文字具體性作為是判別重要性的依據之一。
本研究的實驗結果在未來可以提供商品開發人員在進行未來產品開發時有個更好的參考依據，並同時以自動化的方式分析使用這對於產品的輿情，能夠在有限的開發周期內擬定功能開發的順序，產出完好的設計取向!

關鍵字

線上評論 ； 情緒分析 ； 自然語言處理 ； 產品開發設計 ； 機器學習

並列摘要

This research mainly explores how the emotional distribution of online reviews affects the future design trends of products. In the age of Internet maturity, shopping on online platforms is the daily life of many people, At the same time, under the wave of big data, how to obtain various information of consumers from the numerous data on the internet platform has become a very popular research field in online consumer behavior research. The smartphone market discussed by the research is a market with fierce competition and relatively short product life cycles, How to find design trends that impress consumers in a relatively short product life cycle is a very important topic for smartphone manufacturers. As far as we know, most of the current research in most related fields is to explore the sentiment analysis and scoring prediction of consumer reviews, and seldom focus on emotion and product design. On the other hand, when considering the importance of consumer reviews, this experiment uses Construal-level theory and previous research as a reference, and the concreteness of the text is used as one of the basis for judging the importance. This research aims to through such research, we can use automated methods to find the potential relationship between emotions and product design, and use it as a guideline for future product development. The main motivation of this research is to find out consumers' emotions about the product through the review left by consumers on the online platform and the exchange of opinions with each other. Find out consumers' emotions about the product, and analyze how the distribution of emotions in the reviews affects the future design trends of the product. This experiment obtained two websites from Amazon and Gsmarena as the source of consumer reviews, the review type of these two target websites is a consumer-to-consumer forum, compared with the reviews of professional reviewers, it is more able to obtain the feedback and emotions of most consumers for the product. In the part of emotion analysis, our research obtained 20,000 tweets as the training data set, and after comparing with other models, the BERT-base model was selected as the emotion analysis model.

From the research results, we can find that in the U.S. smartphone market, those angry reviews have a more obvious impact on the brand’s market share than other emotions. On the other hand, after comparing the product features of the previous generations, our research found that angry reviews have a significant impact on the future design direction of the product. The product features mentioned intensively in the angry reviews will be significantly upgraded or improved in the next generation, In the emotional distribution of these product features, the proportion of anger is also significantly reduced.

From the research results, we can infer that angry emotions are more relevant to the future design of the product than other emotions in the reviews of smartphones, at the same time, it also has a significant impact on the market share of the product, The results of this research can provide future smartphone developers with many opinions and emotions based on online reviews, to provide the decision-making assistance of the development project of the next generation product in a short development time, and win the favor of consumers.

並列關鍵字

Online Review ； Emotional Analysis ； Product Design ； BERT ； Nature Language Processing

參考文獻

Ady, M. and D. Quadri-Felitti (2015). "Consumer research identifies how to present travel review content for more bookings." Retrieved form http://webcache. googleusercontent. com/search.

Bakshi, R. K., et al. (2016). Opinion mining and sentiment analysis. 2016 3rd International Conference on Computing for Sustainable Global Development (INDIACom), IEEE.

Brysbaert, M., et al. (2014). "Concreteness ratings for 40 thousand generally known English word lemmas." Behavior research methods 46(3): 904-911.