11 Graph-based natural language processing

 

This chapter covers

  • A simple approach to decompose a text and store it in a graph
  • How to extract the hidden structure of unstructured data via natural language processing
  • An advanced graph model for taming text

Let’s start this new topic by considering the most common applications that deal with natural language (in different format) for providing services to end users. You likely use them every day, probably without even noticing how complex and useful they are.

Chapter 4 dealt with text to implement a recommendation engine that uses the content related to the items, such as the description of a product or a movie plot. In that case, this data was used to compare items or user profiles, find commonalities (specifically, similarities) among users or items, and use them to suggest something that might be of interest to the current user. Figure 11.1 presents the high-level structure of a content-based recommendation engine taken from chapter 4.

Figure 11.1 A content-based recommendation engine, as presented in chapter 4
CH11_F01_Negro

The item analyzer and user profiles builder deal with text to make it available during the recommendation phase. The result of their analysis is stored in such a way that is it easy to access and query during the model generation and the recommendation process.

11.1 A basic approach: Store and access sequence of words

11.1.1 Advantages of the graph approach

11.2 NLP and graphs

11.2.1 Advantages of the graph approach

Summary

References