Working with SQL and graph databases

Digital text editions, speech databases or video repositories contain millions of interconnected data items. For example, text archives contain documents which in turn are structured texts which consist of parts, chapters, sections, etc. with particular linguistic properties, historical or other relationships.

Databases were developed for huge amounts of data. They provide secure storage, a quick data retrieval via standardized query languages, and multi-user access. Furthermore, they feature programming language interfaces, so that they can be used either from standard application programs such as Excel, R or SPSS, or from programming languages such as perl, Python, Java etc.

The relational data model views the world as interconnected tables which can be accessed and manipulated via simple relational operators. The query language SQL has become the de facto standard language for relational databases -- its English-like query language make it easy to learn, its declarative nature allows one to specify what the result should be rather than how to compute it.

Although SQL and the relational model are still dominant, new data models have been developed which promise to be better suited for modern demands. A particularly interesting new approach is that of graph databases: the world is seen as nodes with properties interconnected via named arcs. The query language Cypher is a quasi-graphical language, and queries are simply graph representations of the results to extract from the database.

In this course I will develop a database for speech databases. This database will first be implemented as relational tables in SQL, and we will learn to express complex queries in SQL. In the second half of the course the database will be remodelled as a graph database - at the end of the course you will be able to critically evaluate the two data models and make an informed decision on which type of database to use for a given application.