Database Reference
In-Depth Information
TFIDF, data analysts can quickly go through the reviews and identify what aspects
are perceived to make bPhone a good product or a bad product.
Figure 9.12
Reviews highlighted by TFIDF values
Topic models such as LDA can categorize the reviews into topics.
Figures 9.13
and
9.14
show circular graphs of topics as results of the LDA. These figures are
produced with tools and technologies such as Python, NoSQL, and D3.js.
Figure
9.13
visualizes ten topics built from the five-star reviews. Each topic focuses on
a different aspect that can characterize the reviews. The disc size represents the
weight of a word. In an interactive environment, hovering the mouse over a topic
displays the full words and their corresponding weights.
