Can digital transformation help to understand financial ...
Transcript of Can digital transformation help to understand financial ...
Can digital transformation help to understand financial
inclusion?: The potential of Big Data and analytics
Madrid, October 2018
6th IFFM Annual Meeting
Noelia Cámara, BBVA Research
Brussels, 6 November 2018
“Financial inclusion is at a turning point. Due to advances in technology, the
unprecedented advent of transactional and behavioral big data and greater
multistakeholder collaboration, there is a realistic opportunity to reach the financially
excluded – estimated to be 2 billion – and the many more who are underserved.”
World Economic Forum. White paper, January 2018
The role of public Big Data as an additional source of information for analyses (policymakers and researchers)
Text minning insights: tools and databases
• Information is released subject to
editor’s criteria limited news are
published, need to prioritize
• Relevance of the news: coverage
• Sentiment analysis: compute the tone of
each topic, platers and geographies in
the dialogue
• Software to exploit information:
BigQuery and R or Python
• There is not filters for publishing
information
• Short-run trends, more immediate
• Topic analysis: with a data driven
process, we identify the most
important topics
• Software to exploit information: R
or Python
Text is the new data. Different sources of text information offer different perspectives
Social Media: TwitterMedia: GDELT
• The Global Database of Events, Language and Tone (GDELT) Project is a real-time global open
database of human society according to the world’s news media, reaching deep into local events,
reactions and emotions of every place of the world in near-real time.
• The GDELT Project monitors every accessible print, broadcast, and online news report around the
globe every 15 minuites in over 100 languages. Information is processed using a vast pipeline of
algorithms to identify thousands of emotions (from anxiety to happiness), millions of narrative
themes (from women's rights to clean water access), as well as locations, people, organizations, and
other indicators.
• We exploit media information (GDELT) through Big Data techniques using Natural Language
Processing (NLP) and sentiment analysis to measure the extent of Financial inclusion related
themes coverage and their perception on the media across the world and over time
• This analysis reflects how countries, institutions, societies and Governments stand on Financial
inclusion related topics and gives a comprehensive view of the main related topics as well as of
people and organizations that play a role in the financial inclusion dialogue
Text minning for getting insights on financial inclusion
related themes: Mass media
GDELT Project is a real-time global open database
The potential of Big Data and analytics: Tone and Coverage
Media coverage over time 2015-2018 (Mov avg 90
days. Relative ratio with respect to total news)
Media sentiment over time 2015-2018 (Mov avg 90 days)
Sentiment: once each news piece is translated into English, GDELT applies more than 40 different dictionaries that
classify words associated with positive and negative tone as to compute the average “tone” of all documents containing
one or more mentions to the events we are looking for. Common values for the score range between -10 (negative) and
+10 (positive), with 0 indicating neutral.
The potential of Big Data and analytics: Language
Media sentiment for financial inclusion over time
2015-2018 (Mov avg 90 days)
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
All sources Only LATAM
LatAm vs. rest of the world
• Language of news could be a
good strategy for getting
insights on the sentiment by
geography
• There is a more positive tone in
LatAm that decreases in time
and converges to the one in the
rest of the world
Sentiment and Coverage by country: Financial Inclsuion
Main related topics with financial inclusion in the media
Most outstanding Financial inclusion-related
topics in the media(2015-2018. Most commented topics with Financial Inclusion
based on media coverage)
Evolution of the most outstanding Financial
inclusion-related topics: by field(2015-2018. Media coverage -size- and sentiment –color)
The treemap represents the distribution of the number of news related to fintech by
topic. Rectangles size shows the share of media coverage of each topic.
Short term trends in financial inclusion: social networks
information
Web: specialized sources
• Natural language understanding technologies to
developers, including sentiment analysis, entity
analysis, entity sentiment analysis, content
classification, and syntax analysis
• Network analysis: analysis of the relationship
between: topics, geographies and players
• Software: NLP Google API (ML, Tensorflow)
Web: specialized sources
Thank you ☺
Sentiment and Coverage by country: Financial Inclusion
Media coverage over time 2015-2018 (Mov avg 90
days. Relative ratio with respect to total news)
Media sentiment over time 2015-2018 (Mov avg 90 days)
Media sentiment evolution over the world