Improving the accuracy and diversity of feature extraction from online reviews using keyword embedding and two clustering methods

In product design, it is essential to understand customer's requirements for product specifications. Traditional methods including surveys and interviews are still widely used to solve this problem, but with the increase of online channels such as Twitter and YouTube, customer opinions that can be collected online have increased exponentially. This online data has the advantage that it can be collected faster and cheaper than traditional surveys. Naturally, many studies have been conducted to analyze customer opinions on product design using online data. Among them, this thesis focused on the word embedding and clustering which is an automated feature extraction method using online product review data. The methodology does identify product features but has some limits. The research presented in this thesis addresses those limitations and proposes a new methodology to solve them. The improved results of the proposed methodology are demonstrated in case studies for three categories of products.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Industrial Engineering

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2020

Author dc:creator

• Park, Seyoung

Contributors dc:contributor

• Kim, Harrison M

• Online review, word embedding, word clustering, feature extraction

Statement dc:rights

• Copyright 2020 Seyoung Park

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/108292

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/108292