A simple web application that is talking to the graph database Blazegraph. The point of this tutorial is to provide anyone versed in RDF the ability to run a graph database and write a simple web application over it.
Knowledge graphs are part of symbolic AI. In short they capture what humans mean when they use certain words e.g. when you search for "Jaguar", do you mean the animal or the car brand. For example, at Zalando, the workplace of the author, understanding the customers' language in search and talking to customers in a language they understand is very important, and the Fashion Knowledge Graph build at Zalando drives that.
This tutorial will help you to set up our own graph databases, add data into it, and use it in a web application. It will give you a quick start to building and hosting your own knowledge graph. If you want to explore the topic further in some of your applications, you can use this technology and expand on it.
This tutorial does not teach the basics of Resource Description Framework (RDF), Turtle syntax, and SPARQL syntax. It is also not explaining the basics of JavaScript syntax. Some adequacy in all these languages is required. But you do not need to know these languages to go through this tutorial, since everything is written out for you in the form of examples.
Make sure you have installed Docker
Make sure you have Node.js installed and npm that should come with it. You can check this by running node -v and npm -v )
Either download or git clone this repository to your local machine.
Open up a terminal and navigate to this repository.
For Mac users
If you downloaded this repository, most likely a ZIP file is in your
Downloads folder. Unzip the file and then navigate to
the folder, for example with:
cd Downloads/graph-application-master/
For this tutorial I am using Blazegraph graph database, because it is open-source and most-likely Amazon Neptune available on AWS is based on Blazegraph. Certainly the experience of using both is the same.
The easiest way to run Blazegraph is to use a popular Docker image. You can find those in Dockerhub. In this tutorial we will use lyrasis/blazegraph.
Make sure you have installed Docker
Then run the following command to run Blazegraph according to the chosen docker image.
docker run --name blazegraph -d -p 8889:8080 lyrasis/blazegraph:2.1.4
To see the dashboard for Blazegraph, navigate with your browser to the localhost port 8889.
Make sure you have Blazegraph running. You can see that it is running by typing
docker ps
Navigate to the dashboard of Blazegraph localhost port 8889 and choose the tab "Update"
Below the big text edit field, you can spot a button saying "Choose file". This is where you can upload an RDF Data file in several formats (N-Quads, N-Triples, N-Triples-RDR, Notation3, RDF/XML, JSON, TriG, TriX, Turtle, Turtle-RDR). To get you started you can upload the Turtle file on clothing materials found in the data folder of this tutorial.
The contents of the TTL file appears in the empty text box and by clicking "Update" below the contents is uploaded in the graph database.
Congratulations! If you have managed to follow all steps of this tutorial, then you are now running a graph database with some data in it. You can test this data by running a few SPARQL queries for it.
Navigate to the Blazegraph Dashboard and go to the tab that says "Query". In the big text input field add your query and then press "Execute" to run it. Below are some example queries that you can run:
QUERY 1 See all classes defined in your graph
prefix owl: <http://www.w3.org/2002/07/owl#>
prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
SELECT ?class_labels ?iri
WHERE {
?iri rdfs:label ?class_labels ;
rdf:type owl:Class .
}
QUERY 2 See all immediate instances of "clothing material"
prefix owl: <http://www.w3.org/2002/07/owl#>
prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
prefix ns: <http://www.example.org/myexample#>
SELECT ?class_labels ?iri
WHERE {
?iri rdfs:label ?class_labels ;
rdf:type ns:clothing_material .
}
QUERY 3 See all instances including sub-classes of instances of "clothing material"
prefix owl: <http://www.w3.org/2002/07/owl#>
prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
prefix ns: <http://www.example.org/myexample#>
SELECT ?class_labels ?iri
WHERE {
?iri rdfs:label ?class_labels ;
rdfs:subClassOf*/rdf:type ns:clothing_material .
}
You should have now a way to upload and read data to your graph database. Next step is to build a simple web app on top of the data you are storing in your graph database. For this we are using a simple express JavaScript file and one Embedded JavaScript template file to render a simple view with data coming out of the graph.
Navigate to the local folder of this project (it would be called graph-application or graph-application-master).
Requirements: Make sure you have Node.js installed and
npm that should come with it. You can check this by running node -v and npm -v )
Install all the dependencies of this project with:
npm install
Install express:
npm install express --save
Run the index.js file of this repo with:
node index.js
Navigate with your browser to localhost port 3000 and you should see the friendly "Hello World!" greeting. To understand why you are seeing this, look at line 23 in the index.js file.
We are using the Embedded JavaScript templates to render a view and pass some parameters to it. Make sure you have ejs installed:
npm install ejs
Now navigate to http://localhost:3000/graph to see a view that is rendered out of data that comes from the graph.
To understand what is happening in this code, let's go through it:
On line 12 of the index.js file, we are defining QUERY3 from
our examples above into the variable query.
In the lines 25-37 of the index.js file, we are defining what needs to be done when '/graph' is requested in the URL.
On line 26 we are passing the query variable as part of a GET Request that calls the Blazegraph SPARQL endpoint
as defined in line 8 and asks for the response to be in JSON format {headers: {"Accept":"application/sparql-results+json"}}.
From that response we are extracting the iri and labels values that should be returned from our query. These values are then
pushed to the variable results. These results are read by the graph.ejs template and rendered as a view.
Navigate to the root of this repository. Run:
docker-compose up
Then navigate to http://localhost:3001/graph to see the end result.