Search for Public Conversations

Strategic decision making with exploratory search
Toby Mostyn
CTO Polecat
Agenda
What is the point of Polecat?
Failing to meet the information need
Queries: Handling complex topics
Results: Finding “insights” in the noise
Solving both problems: an exploratory paradigm
Darwinian algorithms
What is the point of Polecat?
Intelligent searching
on
public conversations
Unlocking the Potential of Social Media!
Architecture
News
Social media
Importer
Blogs
Information
Extraction
Indexing
Search
platform
MeaningMine
What is the point of Polecat?
Failing to meet the information need
Queries: Handling complex topics
Results: Finding “insights” in the noise
Solving both problems: an exploratory paradigm
Darwinian algorithms
Failing to meet the information need
Forming Policy
Brand Management
What are the issues that
people care about most?
What/who is my product
associated with?
Overview
Issue Management
Briefing
Give me an up to the minute
/ long-term info on an issue
I need to know,quickly,all
about x
Beyond traditional search
Irish Government: setting the agenda for the Irish Economic Forum
Query + results = failure to meet information need
What is the point of Polecat?
Failing to meet the information need
Queries: Handling complex topics
Results: Finding “insights” in the noise
Solving both problems: an exploratory paradigm
Darwinian algorithms
Queries: handling complex topics
Information need:
What is the discussion around innovation in the UK economy?
All (relevant) documents are important!
Simple keyword = failure
User unable to assess and select keywords
User unable to formulate complex boolean query
Queries: handling complex topics
Query by document
 Feed in 1 to n documents
 Pseudo relevance feedback
 Query extraction -> query expansion
Exploratory interface

Results become query prompts

Users build iterative queries
What is the point of Polecat?
Failing to meet the information need
Queries: Handling complex topics
Results: Finding “insights” in the noise
Solving both problems: an exploratory paradigm
Darwinian algorithms
Results: Finding “insights” in the noise
Goal: provide the user with an exploratory overview of the results
Solution: Insights: extracted information/statistics that describe the data
 Information Retrieval Statistics
 Topic models
 Sentiment analysis
 Entity extraction
Show me the data!
Results: Finding “insights” in the noise
What is the point of Polecat?
Failing to meet the information need
Queries: Handling complex topics
Results: Finding “insights” in the noise
Solving both problems: an exploratory paradigm
Darwinian algorithms
Solving both problems: an exploratory paradigm
What is the point of Polecat?
Failing to meet the information need
Queries: Handling complex topics
Results: Finding “insights” in the noise
Solving both problems: an exploratory paradigm
Darwinian algorithms
Darwinian algorithms
Business
Polecat
Ecosystem
Academia
Darwinian algorithms
 Public search application: summarisation engine
 Plug-in architecture for 3rd party algorithms/ visualisations
 Crowd source judgements
 Published evaluation tables (weekly/monthly)
Darwinian algorithms
Darwinian algorithms
Ranked insight by query type
Ranked insight combinations
Ranked visualisation by insight type
Individual scores for each contributor