Cheuk Ting Ho
Developer advocate / Data Scientist - support open-source and building the community.
Are you supporting the right politician?
Graph Visualization of Voting Data
Cheuk Ting Ho
Grab the slides:
http://bit.ly/CivicLabGraphViz
About me
Co-organizer of
Open Source contribution
Creator of
Volenteer of
Developer Advocate of
Have you ever check what issues that a politician support before you vote?
Or, do you just vote for them because of the party that they are in?
Do you think your Councilor represent you in local issues?
Have you ever check what they vote for in the Dublin Council?
Politics can be complicated...
But data never lies...
(only people who interpete them does)
I gonna teach you how to analyze the voting data
So...
What's a Graph?
And Why?
-
much easier to interpret the model
-
understand how it maps to real world entities compared to the relational example
SELECT Name from TABLE where Person_ID = (SELECT mother from TABLE where Name="John")
SELECT Name from TABLE where Person_ID = (SELECT mother from TABLE WHERE Person_ID = (SELECT mother from TABLE where Name="John"))WOQL.and(
WOQL.triple("v:Person", "mother", "v:MotherID"),
WOQL.triple("v:MotherID", "name", "v:MotherName"),
WOQL.triple("v:MotherID", "mother", "v:GrandmotherID"),
WOQL.triple("v:GrandmotherID", "name", "v:GrandmotherName"),
)Getting
the Data
Install TerminusDB
Setup Checklist
Do you have TerminusDB installed?
Yes
No
Good to go
Do you have Docker?
Yes
No
Which OS?
Mac / Linux / Window 10 Pro
No
Others
Do you have TerminusDB installed?
Create a Graph DB
1. Make sure TerminusDB is running
2. http://localhost:6363/console
3. Create a new Database
Create a Schema
WOQL.when(true).and(
WOQL.doctype("Party")
.label("Party")
.description("Political Party"),
WOQL.doctype("Representative")
.label("Representative")
.description("An elected member of the Dublin city council")
.property("member_of", "Party")
.label("Member of").cardinality(1),
WOQL.doctype("Similarity")
.label("Similarity")
.property("similarity", "decimal")
.label("Similarity")
.property("similar_to", "Representative")
.label("Similar To").cardinality(2)
)Load-in
the Data
let csv = WOQL.get(
WOQL.as("councillor_a","v:Rep_A")
.as("councillor_b", "v:Rep_B")
.as("party_a", "v:Party_A")
.as("party_b", "v:Party_B")
.as("distance", "v:Distance")
).remote("https://terminusdb.com/t/data/council/weighted_similarity.csv");
let wrangles = [
WOQL.idgen("doc:Party", ["v:Party_A"], "v:Party_A_ID"),
WOQL.idgen("doc:Party", ["v:Party_B"], "v:Party_B_ID"),
WOQL.idgen("doc:Representative", ["v:Rep_A"], "v:Rep_A_ID"),
WOQL.idgen("doc:Representative", ["v:Rep_B"], "v:Rep_B_ID"),
WOQL.typecast("v:Distance", "xsd:decimal", "v:Similarity"),
WOQL.idgen("doc:Similarity", ["v:Rep_A", "v:Rep_B"], "v:Rel_ID"),
WOQL.concat("v:Rep_A similarity v:Distance to v:Rep_B", "v:Rel_Label")
];
let inputs = WOQL.and(csv, ...wrangles);let inserts = WOQL.and(
WOQL.insert("v:Party_A_ID", "Party")
.label("v:Party_A"),
WOQL.insert("v:Party_B_ID", "Party")
.label("v:Party_B"),
WOQL.insert("v:Rep_A_ID", "Representative")
.label("v:Rep_A")
.property("member_of", "v:Party_A_ID"),
WOQL.insert("v:Rep_B_ID", "Representative")
.label("v:Rep_B")
.property("member_of", "v:Party_B_ID"),
WOQL.insert("v:Rel_ID", "Similarity")
.label("v:Rel_Label")
.property("similar_to", "v:Rep_A_ID")
.property("similar_to", "v:Rep_B_ID")
.property("similarity", "v:Similarity")
);
WOQL.when(inputs, inserts)Making Sense of the Data
/*******************
Query Script
********************/
WOQL.limit(1000).and(
WOQL.triple("v:Subject","similar_to","v:Value"),
WOQL.triple("v:Subject","similar_to","v:Value2"),
WOQL.triple("v:Subject","similarity","v:Similarity"),
WOQL.triple("v:Value","member_of","v:Party"),
WOQL.triple("v:Value2","member_of","v:Party2"),
WOQL.not().eq("v:Value","v:Value2"),
WOQL.opt().triple("v:Value2","label","v:Lab2"),
WOQL.opt().triple("v:Value","label","v:Lab1"),
WOQL.eval(WOQL.divide(1, WOQL.exp("v:Similarity", 4)), "v:Distance")
)
/***************************************
View Script for Dublin City Council
****************************************/
view = View.graph();
view.height(800);
view.node("v:Subject", "v:Lab2", "v:Lab1", "v:Party2", "v:Party", "v:Similarity", "v:Distance").hidden(true)
view.node("v:Similarity").hidden(true)
view.edge("v:Value", "v:Value2").distance("v:Distance").text("v:Distance").weight(0.04)
view.node("v:Value").text("v:Lab1").icon({ label: true})
view.node("v:Value2").text("v:Lab2").icon({ label: true})
view.node("v:Value", "v:Value2").charge(-4999).collisionRadius(30)
view.node("v:Value").v("v:Party").in("doc:PartySolidarity").color([5, 25, 22])
view.node("v:Value2").v("v:Party2").in("doc:PartySolidarity").color([5, 25, 22])
view.node("v:Value").v("v:Party").in("doc:PartySocial%20Democrats").color([25, 25, 225])
view.node("v:Value2").v("v:Party2").in("doc:PartySocial%20Democrats").color([25, 25, 225])
view.node("v:Value").v("v:Party").in("doc:PartySinn%20F%C3%A9in").color([25, 225, 25])
view.node("v:Value2").v("v:Party2").in("doc:PartySinn%20F%C3%A9in").color([25, 225, 25])
view.node("v:Value").v("v:Party").in("doc:PartyLabour%20Party").color([255, 0, 0])
view.node("v:Value2").v("v:Party2").in("doc:PartyLabour%20Party").color([255, 0, 0])
view.node("v:Value2").v("v:Party2").in("doc:PartyFine%20Gael").color([0, 0, 255])
view.node("v:Value").v("v:Party").in("doc:PartyFine%20Gael").color([0, 0, 255])
view.node("v:Value").v("v:Party").in("doc:PartyFianna%20F%C3%A1il").color([100, 200, 100])
view.node("v:Value2").v("v:Party2").in("doc:PartyFianna%20F%C3%A1il").color([100, 200, 100])
view.node("v:Value").v("v:Party").in("doc:PartyGreen%20Party").color([25, 225, 125])
view.node("v:Value2").v("v:Party2").in("doc:PartyGreen%20Party").color([225, 225, 125])
view.node("v:Value").v("v:Party").in("doc:PartyIndependent").color([25, 25, 25])
view.node("v:Value2").v("v:Party2").in("doc:PartyIndependent").color([25, 25, 25])
view.node("v:Value").v("v:Party").in("doc:PartyPeople%20Before%20Profit").color([225, 25, 25])
view.node("v:Value2").v("v:Party2").in("doc:PartyPeople%20Before%20Profit").color([225, 25, 25])
view.node("v:Value").v("v:Party").in("doc:PartyWorkers'%20Party").color([225, 225, 225])
view.node("v:Value2").v("v:Party2").in("doc:PartyWorkers'%20Party").color([225, 225, 225])
So How About UK Parliament
You time to shine
🌟
And win some prices
Competition deadline: 17th Mar 2020 (St. Patrick's Day!)
Team: 1-3 ppl
Format: a Jupyter notebook (.ipynb file) or a blog post
*Submission need to be publicly available (e.g. host on GitHub or published blog post)
Show us any interesting findings with the UK Parliament data using a similar method. Be creative!
Originality, Technical correctness, Informative
Prices:
1. £100 amazon gift card for the winning team
2. each team member will get a TerminusDB swag bundle
(include Hoodie Jacket, E-Coffee Mug and much more)
Winner will also be invited to speak at our meetup in Dublin
(and/or London!)
Submittion/ Questions: luke@datachemist.com
Name(s), contact email(s), link to submittion
Webinar:
Anatomy of a Knowledge Graph
10th March - 5pmCET
To get the newest update👍:
Follow us on Twitter: @TerminusDB
Website: https://terminusdb.com/
Join the community:
Civic Lab Summit - Are you supporting the right politician?
By Cheuk Ting Ho
Civic Lab Summit - Are you supporting the right politician?
Graph Visulization on Voting Data
