Text Analytics & Sentiment Analysis - MathWorks · 7 What is Text Analytics/NLP? Artificial...

Post on 19-Jun-2020

14 views 2 download

transcript

1© 2019 The MathWorks, Inc.

Text Analytics & Sentiment Analysis

Alex Link, Application Engineer

2

3

4

5

Big Picture

Topic Modeling

Un

str

uctu

red

Data

Str

uctu

red

Data

Social

Media

Database

In-house

Spreadsheets

Traditional Data

Analytics

Decision

Reports

Text Analytics

6

Where is this data?

Data Sources for Big Data Analysis

7

What is Text Analytics/NLP?

Artificial Intelligence

Machine

Learning

Deep Learning

Natural

Language

Processing

Text Analytics

= Natural Language Processing + Machine Learning

8

Why now?

“About 80% of data of an organization processes daily is unstructured data” – Gartner 2018

60 to 80%

growth per year

9

What is text analytics being used for today?

Key Applications

Sentiment Analysis

Topic Modeling

Text Classification

Text Generation

Voice of Customers

Market/Competitive Intelligence

Fraud/Legal Risk Detection

Compliance Check

Triage/Routing

Conversational Agents

10

Sentiment Analysis

Other Applications:

• Automating the classification of reviews, whether positive or negative

• Analyzing surveys to understand why customers are satisfied or dissatisfied

• Assessing counterparty credit risk

Positive (+)(growth, advances,

up, strong)

Neutral Negative (-)(bust, difficulty,

lack, struggle)

Goal: Determining real-time sentiment scores for use in financial trading

strategies

Buy! Sell!

Hold

11

Integrate

Deployment

(Desktop, Web, Enterprise)

Share and Integrate

Scale

Text Analytics Workflow

Raw Data

Preprocess

Clean Data

Access Text

Text, PDF, Word, Excel,

HTML

Visualize

Languages

Develop Predictive

Models

Statistics

Deep Learning

Machine Learning