원문정보
초록
영어
Advances in Internet and Smartphones are generating huge amount of customers’ opinion (or VOC: voice of customer) in the online world. VOC is very useful for marketing or product design departments to see the customers’ opinion for the products. In this paper, VOC in online review sites has been collected, stored and analyzed in order to find out what kind of factors can affect the customer preference for the electronics products. We have implemented a VOC analysis system that includes ETL (extract, transformation, and loading), natural language processing, positive/negative analysis, and result visualization. Sensitivity adjectives and synonyms are collected and stored as a database for more correct sentiment analysis. Real experiment for VOC data with 39 Giga bytes from well-known review sites in Korea shows that the correctness for the classification of the positive/negative sentences and the causes of the positive/negative are 73% and 70%, respectively. The evaluation has been done based on the manual inspection for the analysis result
목차
1. Introduction
2. Related Work
3. System Architecture and VOC Data Analysis
3.1. System Architecture
3.2. Data Collection and Storing
3.3. Filtering, Cleansing, and Natural Processing
3.4. Sentiment Analysis and Visualization
4. Experiment
4.1. System Environments
4.2. Frequency Analysis
4.3. Sentiment Analysis
4.4. Evaluation
5. Conclusions
References