Cognitive Intelligence in Relational Databases

Thumbnail Image

Files

Athley_umbc_0434M_11692.pdf (3.26 MB)

Links to Files

Permanent Link

http://hdl.handle.net/11603/15494

Collections

UMBC Theses and Dissertations

Author/Creator

Author/Creator ORCID

Date

2017-01-01

Type of Work

theses
Text

Department

Computer Science and Electrical Engineering

Program

Computer Science

Citation of Original Publication

Rights

This item may be protected under Title 17 of the U.S. Copyright Law. It is made available by UMBC for non-commercial research and education. For permission to publish or reproduce, please see http://aok.lib.umbc.edu/specoll/repro.php or contact Special Collections at speccoll(at)umbc.edu
Distribution Rights granted to UMBC by the author.

Subjects

cognitive querying
databases
word2vec
Word embedding

Abstract

We evaluate the applicability of distributed language embedding techniques from the domain of natural language processing to relational data. Relational data is typically stored in SQL databases. We apply modern distributed representations of words (Tomas Mikolov 2013c) and paragraph (Quoc V. Le 2014) techniques to this structured data and attempt to unlock the potential of enhanced cognitive querying. The research intention is to be able to perform queries which are non-trivial to perform using the SQL dialect alone. We tokenize the IMDB 5000 movie data-set to generate embeddings using word2vec and a modified version of doc2vec that we term as row2vec. We discuss the effects of various hyperparameter choices and tokenization techniques. We visualise these embedding using PCA and present the results for certain queries. Keywords: Word embedding, databases, word2vec, cognitive querying.