Database Reference
In-Depth Information
Summary
This chapter has discussed several subtasks of text analysis, including parsing,
search and retrieval, and text mining. With a brand management example, the
chapter talks about a typical text analysis process: (1) collecting raw text, (2)
representing text, (3) using TFIDF to compute the usefulness of each word in the
texts, (4) categorizing documents by topics using topic modeling, (5) sentiment
analysis, and (6) gaining greater insights.
Overall text analysis is no trivial task. Corresponding to the Data Analytic Lifecycle,
the most time-consuming parts of a text analysis project often are not performing
the statistics or implementing algorithms. Chances are the team would spend most
of the time formulating the problem, getting the data, and preparing the data.
