Sentiment Analysis

Inspired by a very poorly presented Survey, I stared to delve into how to better present the data as a pitch for that potential client. As per usual, this raised more questions then it answered and I started to look at the detailed workflow to process survey results and present them.

What quickly became obvious was there was a need to augment the survey outcome data by tying it back to the original quest for knowledge/insight, participants reasons (dare I say motivation)
for engagement in this quest and other sources of data to better elicit knowledge/insight.

Not so quickly came my own insight into this process, that there are three main drivers for this engagement;

Future States
Future States
Directed 
[Not supported by viewer]
Organic
Organic
answers
+ time
+id
+location
[Not supported by viewer]
fill in
fill in
interest
interest
anounciates
anounciates
Survey
Participant
Survey<br>Participant
provides
questions
[Not supported by viewer]
drives
drives
Required Knowledge
Required Knowledge
feed
feed
Survey
Results
[Not supported by viewer]
Anonymise
Filter
Junk/Abuse
[Not supported by viewer]
Questions/
Advertising
[Not supported by viewer]
Privacy Rules
Name
Location
Phone
Email, etc
[Not supported by viewer]

Cleaned Survey Results
<br>Cleaned Survey Results<br>
NormaliseText
Stem Words
Contractions
ASCII Filter
Spell Check
[Not supported by viewer]
filter by tag
filter by tag
Parts of Speech (POS) Tagging
[Not supported by viewer]
link words
link words
Shallow Parse
Noun Phrase
Ad/Verb Phrase
Adh Phrase
Prepos. Phrase

[Not supported by viewer]
Graph Network Visualisation
[Not supported by viewer]
Constituency parser 
Dependency Parsing
Named Entity Recognition
[Not supported by viewer]
Sentiment Analysis
Polarity Score
Neg[-1],Pos[+1]
SubjectivityScore
Subj[0], Obj [+1]

[Not supported by viewer]

<br style=”font-size: 12px”>
Word Cloud
Word Cloud<br>
Full Text
Full Text
Public Text
Public Text
defines
defines
Survey
Designer/
Owner
[Not supported by viewer]
Thus, a sentence typically follows a hierarchical structure consisting the following components,
sentence → clauses → phrases → words
[Not supported by viewer]
rank words
rank words
metadata
metadata
metadata
metadata
Tags
Tags
Named Entities
#tags
platform
[Not supported by viewer]
posts
posts
writes
writes
1
1
2
2
4
4
3
3
Quad Polarity,Subj.
1. Optimist 1,1
2. Pessimist -1,1
3. Cynic -1,0
4. Dreamer 1,0
[Not supported by viewer]
Survey Sentiment Workflow
Survey Sentiment Workflow
feeds
[Not supported by viewer]
API
API
drives
drives
ellicits
ellicits
Shared
Aspiration
Collaborative
Future
Vision/Values
[Not supported by viewer]
feed
feed
TragectoRate
TragectoRate<br>

Social Media Platform
 Data
[Not supported by viewer]


Snapshot of  Relevant Data
[Not supported by viewer]
Other data Sources
[Not supported by viewer]
  • Satellite/Terrain Data
  • Census/Demographic Data
  • Resource/Vege/Water Data
  • Cadastral/Property Data
  • Streetmap Data
  • Transport/Service Data
  • Manufacturing/Consumption
  • Sensor Data
  • Content/Creative/Research
  • Webdata (ip tracking)
  • Shadow/Machine Profiles
[Not supported by viewer]

The worklow consists of;

  • Cleanse and Aggregate data sources to a publishable raw data set.
  • Process data to extract features for presentation.
  • Presentation layer that is relatable to the initial inquiry.