Loading...

Determining Informative Content based on Customer Needs Using Deep Learning on Customer Reviews

Shafie, Amir Hossein | 2020

392 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 53282 (44)
  4. University: Sharif University of Technology
  5. Department: Management and Economics
  6. Advisor(s): Najmi, Manoochehr
  7. Abstract:
  8. Right from the start, customer need has been the center of focus for every firm. It’s hard to find a single firm which is not focused and customer centric. Firms are spending millions of dollars every year to make customer surveys and interviews to find out customer needs, new product development and brand perceptions. After the invention of the internet and the rise of home computers people started to act differently. Most of the purchases went online and customer reviews flooded over the internet. They freely express their mind on websites which is a rich resource for market researchers and marketing firms. UGC is rich enough to analyze customer’s thoughts and needs. The volume of UGC has been generated is overwhelming, therefore the main problem of analyzing UGC rises when we face scale. On scale the majority of UGC contains uninformative content. As a solution to this problem we introduce a model based on artificial intelligence. According to the power of deep learning networks we use convolutional neural networks as our model. A major part of neural network performances relies on training sets. With the help of marketing specialists all over the world, a set consisting of 2300 iPad customer reviews were developed and labeled. Finally, the model performance surpasses eighty percent with the F1-Score of ninety percent on identifying informative content. Although The model's training set is less than the previous models, the performance, precision and F1-Score has raised
  9. Keywords:
  10. Artificial Intelligence ; Machine Learning ; Deep Learning ; Customer Needs ; Big Data ; Marketing ; Opinion Mining ; User Generated Content

 Digital Object List

 Bookmark

No TOC