Simple Graph Database - A Hexastore based Triple Store

---

Developed to help learn the Golang language, this project explores a simple form of graph database, the Triple Store, by implementing the Hexastore architecture from the paper "Hexastore: Sextuple Indexing for Semantic Web Data Management". The project also explores the SPARQL graph query language, by implementing a simplified version of it which I dubbed simplesparql.

Introduction - Usage

As a simple demo, we can scrape your Twitter network and query it with this library. To do so:

1. git clone this library to your designated Go workspace
2. Go into the scripts/ directory
3. Install script dependencies with pip install -r requirements.txt (consider using virtualenv).
4. Run ./create_twitter_followers_graph_example_db.sh. Note: If you don't have a Twitter account, skip this step and a default Twitter network (mine) will be used.
5. Go into the /examples/your_twitter_graph/ folder and run go build
6. A binary is created, which you can run with ./your_twitter_graph
7. Query the n etwork with simplesparql syntax (See the README.md section on it below).

Example starting query

Get all the people that you follow with:

SELECT ?x WHERE { '<YOUR SCREEN NAME>' 'follows' ?x }

Usage - Golang Package

1 .Add to project:

go get github.com/thundergolfer/simplegraphdb or glide get github.com/thundergolfer/simplegraphdb

2. Add to imports:

import (
  "github.com/thundergolfer/simplegraphdb"
)

Package Interface

InitHexastoreFromJSONRows(filename string) (*HexastoreDB, error)

You can setup a Hexastore by passing a filepath to a .json file with the following format:

{"subject": <STRING>, "prop": <STRING>, "object": <STRING>}
{"subject": <STRING>, "prop": <STRING>, "object": <STRING>}
{"subject": <STRING>, "prop": <STRING>, "object": <STRING>}
...
...

InitHexastoreFromTurtle(dbFilePath string) (Hexastore, error)

BETA: Turtle files are currently parsed with d4l3k/turtle, which is quite limited.

Turtle (Terse RDF Triple Language) is a syntax for describing RDF semantic web graphs. You can load an RDF graph specified in turtle syntax with this function.

RunQuery(query string, store Hexastore) (string, error)

Run a well-formed simplesparql query (see more below) against a Hexastore instance. Just returns a printable table of results like:

?target                | ?prop                  |
------------------------------------------------
Cow                    | follows                |
Apple                  | follows                |

SaveToJSONRows(filename string, store Hexastore) error

Not yet implemented, but coming soon

---

simplesparql

simplesparql is an implementation of a subset of the SPARQL query language. It supports the asking of basic questions about a graph, questions like "Who likes 'Manchester United'?", but not anything involving aggregations, grouping, conditionals. With time, these things may be implemented.

The basics of it are that you preface variables with ?, with a variable in the WHERE clause acting like a wildcard (*). For example:

SELECT ?screen_name WHERE { 'jonobelotti_IO' 'follows' ?screen_name }

will return all people that jonobelotti_IO follows (ie. all vertices on the graph connected by a 'follows' edge from the vertice 'jonobelotti_IO'.

You can use up to a maximum of 3 variables in a WHERE clause. All WHERE clauses must have exactly three components. For example:

SELECT ?screen_name, ?property WHERE { 'jonobelotti_IO' ?property ?screen_name }

will essentially return all triples in the graph involving the 'jonobelotti_IO' vertice.