In this webinar Anthony J. Sarkis, Chief Strategy Officer at Parabole, and Steve Sarsfield, VP Product at Cambridge Semantics, explore how portfolio managers are using the recently developed Parabole/ AnzoGraph DB integration as their underlying infrastructure for conducting ML and cognitive analytics at scale to exploit data to identify potential risks and new opportunities.
Determinants of health, dimensions of health, positive health and spectrum of...
Sustainability Investment Research Using Cognitive Analytics
1. Sustainability Investment Research Using
Cognitive Analytics
How Parabole Uses Graph Analytics Database, AnzoGraph DB, to Deliver
Richer and Faster ESG Insights for Portfolio Managers
Anthony J. Sarkis, Chief Strategy Officer, Parabole.ai
Steve Sarsfield, VP Product, Cambridge Semantics
2. Ask us questions.
Slides will be available after
the webinar.
Agenda
• Demo and description of AlphaESG
• A look under the covers –
AnzoGraph DB
3. Training
Data
Training
Models
Linguistic
Techniques
Parabole Learning components
Parabole Text Analysis components
SME designs cognitive analytics
framework (Inputs-Set of Contexts
and their text definitions)
Generated labeled training data
Learning output (white box models)
Semantic
Techniques
Analysis Output
Business solutions
Named Entities
Key Phrases & relationships
Parabole
Business
Soln.
Corpus
Building
Glossary
Content ( long text, short text)Similarity
alphaESG
Financial news analytics
Knowledge Graph (Ontology)
Context Models
Labeled document (training data)
Labeled paragraph (training data)
Labeled sentence (training data)
Labeled phrase (training data)
Word space
Connected graph
Content (text) for analysis
Non-Parabole solutions
AI
Application
AI
Application
Topic models
Regulatory analytics
Risk monitoring
Metamap (Metadata)
Non
Parabole
App
3
API
Full stack cognitive capabilities enable enterprises to
fast-track AI- knowledge mining and deep learning to
extract intelligence from unstructured text
4. 1
2
3
4
Investment
Research
Information Overload
80% of company related data is in textual form
(Alternative Data) and the volume of information is
massive
Information asymmetry
Need to verify whether a corporation has followed
through on disclosed commitments to ESG programs
Third-party Scoring Models
Scoring services provide guidance, however, most
analysts and PM’s need to add their own diligence and
create referenceable files
Research not always aligned with investment strategy
Conducting research aligned to your specific strategies is as
important as using quantitative metrics for investment decisions
ESG Research: Challenges
4
5. Cognitive Models
Parabole
ESG
Analysis
ESG Dashboard (company)
ESG watchlist
News articles,
corporate
Filings, research
reports
Contextualized news
articles, corporate
filings, research reports
ESG contexts and frameworks customized
including SASB guidelines and mapping
Increasing use of non-traditional factors from alternative data makes automation and cognition critical
5
6. Why is alphaESG different?
• Able to process a massive volume of text-based documents with easy
porting to internal news/filing/research feeds
• Format Agnostic (PDF, document, spreadsheet; any text file type)
• Pre-trained models incorporate industry standards and other industry
accepted frameworks (e.g., SASB.org)
• Ability to generate topic signal at differing depths
– topical story level
– company level
• Efficiency
• Customizable inputs:
– Date ranges
– News sources
– Industry/company specific
6
8. Benefits
Topical ESG signals:
– Model will provide a probability of ESG-thematic qualification; generates signal in near real-time
upon introduction to the platform of news stories/topics
Company level ESG signals:
– Find ESG-related stories for the targeted company
– Aggregate and weight by relevance contextual stories for the targeted company
– Aggregated at customizable time intervals (for example, 1,3,6,12 months or daily)
Portfolio monitoring
– a ranking of all portfolio and watchlist companies daily; as new information comes in, alphaESG
updates topic signals
Supervised learning setup
– Training data: news stories + any other required source+ ESG labels
– Industry accepted standards including SASB frameworks
SaaS model (including pre-learnt models and news feed) or on-premises (for large deployments and
custom models)
8
9. Parabole with AnzoGraph
Anzograph DB graph-driven data fabric architecture enables alphaESG to offer
multi-dimensional analytics on diversely formed sustainability data.
ANZOGraph
DB
1
2 3
Harmonization of disjointed and diverse
data through analyzable graph structure
Functional blending of
graph database with an
expansible analytics engine
Standard based query framework
(SPARQL*) and forward-looking open
standard compatibilities (openCypher)
Distributed graph analytics
supporting linear scaling of data
ANZOGraph DB
1
2
3
4
9
11. Rethinking Data Landscape and Connected Data
Centralized
Customer data in a
data warehouse
Follow traditional
data models and
tools
Structured data in
tables
Simple analytics
needs
DATA
Many data
sources
Data
contains
valuable
relation-
ships
Less
structure
Machine
learning and
graph
analytics
Yesterday Today
12. Imagine…
A database technology where you could just store unfettered facts
Susan Johnson is a person
Acme Corp. hired Susan Johnson
Acme Corp. donated to Unicef
Acme
Corp
Susan
Johnson
Unicef
works for
Donates
13. Analytics:
Relationship-centric BI
Core: A graph database
Stores and retrieves data
Has analytical functions
• Relationship queries in the
language(s)
• Easier ontologies for added
intelligence
AnzoGraph DB
Knowledge Graph
Incorporates disparate data
14. Imagine leveraging ontologies to enhance intelligence
Flipper isA Bottlenose Dolphin
Shamu isA Killer Whale
Every Dolphin is an Animal
Every Dolphin is a Mammal (SubClass)
Every Dolphin is in the Delphinidae (Family)
Every Killer Whale is in the Delphinidae (Family)
Bottlenose dolphin is also known as Tursiops truncatus
Flipper and Shamu are both animals
Flipper and Shamu are mammals
Flipper and Shamu are both from the Delphinidae (Family)
Flipper and Shamu are both carnivorous
DATA
INFERENCE
KNOWLEDGE
Leveraging ontologies
What’s the link between Flipper and Shamu?
15. Financial Industry Business Ontology (FIBO)
Entity Types Legal Persons
Formal Organization Corporations
Partnerships Trusts
Ownership Control
By Function Legal Entity ID
Corporate Structure Hierarchies
Susan Johnson’s Job Title is CEO
Acme Corp. hired Susan Johnson
Your facts, augmented
with standard ontologies
like FIBO
+
16. Scale
Imagine billions or even trillions of facts
• LUBM – 110 times faster
than any previous results
• TPC-H (GHIB) – 217 times
faster than a leading OLTP
solution for load and
analysis
• Graph 500 - Load 41.6
million vertices and 1.47
billion edges in 4 ½ Minutes
Massively Parallel – on bare metal, cloud or containers
17. How many new
customers
yesterday/last
month?
What region
sold the most?
What person(s)
are most
influential?
Indirect
connections
between
corporations?
Storing detail
on entities and
how they are
connected.
Viewing
windows of
time and
comparing to
other time
windows.
Standard DW-style
analytics
DetailConnected Analysis Data Science &
Machine Learning
Look for normal
and unusual
behavior.
Data Science for
discovery of new
insight
18. AnzoGraph standard and extra analytical functions
Graph Patterns
Negation
Property Paths
BIND
Aggregates
Basic Federated Query
ORDER BY and offsets
Functions on Strings
Functions on Numerics
Functions on Dates and
Times
Hash Functions
Basic Graph Patterns
Count/Avg
Min/Max
GroupConcat
Sample
Page Rank
Shortest Path
All Path
Label Propagation
Weakly Connected
Components
K neighborhood
Counting Triangles
Inferences (RDFS+)
Labeled Property Graphs
(RDF*)
Window Aggregates
Advanced Grouping Sets
Named Views
Named Queries
Conditional Expressions
User-Defined Extensions
SPARQL 1.1
Standards
AnzoGraph® DB
Extras
Graph Algorithms
and Inferencing
Data Science
Extensions
UDX
Distributions
● Bernoulli
● Binomial
● Chi-squared
● Exponential
● Hypergeometric
● Laplace
● Log Normal
● Logarithmic Series
● Negative Binomial
● Normal
Correlations
● Pearson
Entropy
● Cross Entropy
● Differential Entropy
20. AnzoGraph® DB
Graph Analytical Database
Built for Analytics
Unique Online Analytical
Processing database
engine (OLAP)
Massively
Parallel
The fastest data loading
and analytics capability
Standards-
based
Supports W3C standards
(RDF & SPARQL)
Labelled Property
Graphs under proposed
standard
OpenCypher in 2019
Analytics-rich
Graph algorithms, BI-
style analytics,
inferencing, views, user
defined extensions and
much more.
21. Getting into graph analytics
David Kleiss
david.kleiss@cambridgesemantics.com
https://www.linkedin.com/in/davidkleiss
AlphaESG
https://www.parabole.ai/
Contact Us
Download …or download
Anthony J Sarkis
Anthony@parabole.ai
LinkedIn
AnzoGraph DB
www.anzograph.com