Unified processing of natural language and relational data

This work outlines a method for performing natural language tasks as part of a relational framework. Utilizing features of PostgreSQL as a relational database and its extensibility to allow for word embedding without leaving the relational database. This system can be extended to incorporate several natural language processing (NLP) techniques, such as latent Dirichlet allocations(LDA) or modern models, such as BERT. The combination of NLP and relational operations allows for extracting data from and analyzing text in the same interface used for general data analysis. This combination allows for gathering richer information from existing sources and makes it all available from one standard interface. The declarative nature of SQL allows for more ad-hoc application of NLP techniques. Two case studies using the DBLP dataset demonstrate this integration’s power. Building an LDA model, augmenting the topic labels for greater descriptiveness, and applying preexisting models for semantic analysis.

Name thesis:degree_name

Master of Science (MSc)

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Ontario Institute of Technology

Year dc:date.issued

2022

Author dc:creator

• Stoica, Andrei

Advisors dc:contributor.advisor

• Pu, Ken
• Davoudi, Kourosh

• Query language
• Database
• Natural language processing
• Embedding vectors
• Text processing

Language dc:language.iso

en

Handle dc:identifier.uri

https://hdl.handle.net/10155/1547

OAI identifier oai:identifier

oai:ontariotechu.scholaris.ca:10155/1547