This document provides an overview of document-based research methods and examples of analyzing secondary data sources like social media for business intelligence purposes. It discusses using documents that were created for other purposes, issues to consider, and common public and company-specific data sources. Examples include analyzing the sentiment of tweets about companies; comparing Twitter sentiment to polls; examining language used by an organization and its audience on social media; and identifying emotions and communication styles in social media content.
2. About the presentation
• This presentation has been originally compiled by Jari Jussila,
Häme University of Applied Sciences, @jjussila
• Jukka Huhtamäki, @jnkka, updated the presentation and gave this
lecture at Tampere University TTA-15090 Research Methodology
course during 31 January 2019
• Jari Jussila translated the presentation, and made few updates for
lecture on BBIBP18 Introduction to Business Information
Management at Häme University of Applied Sciences
• Lecture material from Professor Saku Mäkinen (2016) was also
used in compiling this lecture
3. www.hamk.fi
Document based research
• Created (and collected) for different purpose, i.e. secondary data
• Due to increased computational capacity it is more easy to collect
and store documents
• Few things to consider related to document based research
� if documents have been already collected, you save time and effort
� when the documents have been collected for a different purpose, you may
not find an answer to your question
Adapted from Mäkinen (2016)
4. www.hamk.fi
Documents as a source of business
information
- social media
- web pages
- press releases and bulletins
- annual and quarterly reports
- databases
- intranet
- documents
- discussions (audio,
video, chat)
- statistics
- reports
- studies
- commercial statistics
- commercial reports
- studies requiring
membership or
affiliation to access
public not public
Companyspecificgeneral
Adapted from Mäkinen (2016)
5. www.hamk.fi
Common public sources
• Statistics, e.g.
• Statistics Finland: http://www.stat.fi/
• Open Data: https://www.avoindata.fi/fi
• Links to Open Data Sources and Pages: https://avoinhäme.fi/avoin-data-aineisto/
• European Statistics: http://epp.eurostat.ec.europa.eu/
• Finnish Social Science Data Archive: https://www.fsd.uta.fi/
• Public data sources are used in writing a thesis:
• In the introduction section to argument the significance of the topic
• As part of a literature review and theory building
• Or to support empirical material
• Public data sources can be also the main empirical material, when e.g.
• Determining market potential
• Reviewing competition Adapted from Mäkinen (2016)
6. www.hamk.fi
Quarterly Reports
• Text visualization of quarterly
reports as a source of competitive
intelligence (CI)
• Case study of three mobile phone
manufacturers from the years
2000-2001
• Nokia
• Motorola
• Ericsson
Source: Magnusson 2010
7. www.hamk.fi
Impact of Facebook on stock markets
• How Facebook discussions
and activities impact
different investor groups
investing behavior
• Evidence was found that less
professional investors
(households and non-profit
organizations) investing
behavior (purchase of
stocks) is influenced by
Facebook discussions and
activities (e.g. Likes)
Source: Siikanen et al. 2017
8. www.hamk.fi
Event Study
• According to efficient market
hypothesis share prices fully
reflect all available information,
thus it can be observed how
public information influence
share prices
• By calculating on event day (+/-
1 days) how the share price
change was different from
comparison groups market
change we get the estimate of
particular company’s share
price change that is due to
announcement/news/etc.
event
Nokia Corporation share price 14-20 Feb (Source: Nordnet)
Using Secondary Data in Operations Management
Research: Overview and Research Opportunities (Source:
Singhal 2016, p. 25)
9. www.hamk.fi
Data collected with crawlers and scrapers
• Data can be collected from any web
page
• For example, crawler and scraper
implemented to collect data from
Indiegogo crowdfunding platform
• Source code available from:
http://github.com/jukkahuhtamaki/cr
owdfunding-data
• The code must be rewritten. Why
could that be?
Source: Huhtamäki et al. 2015
10. www.hamk.fi
Data collected and analysed from social
media content
Source: https://underhood.co/hamk-university-of-applied-sciences
11. www.hamk.fi
TUT (TUNI) as an example of challenges of
analyzing social media content
https://underhood.co/tampereen-teknillinen-yliopisto-(tty)
12. www.hamk.fi
An analysis of language used by HAMK
and its audience
Source: https://underhood.co/hamk-university-of-applied-sciences
14. www.hamk.fi
Analysis of sentiment and emotions of social
media content
13. Tunnetilojen tunnistaminen
Twitteristä. Jari Jussila, Mika
Boedeker, Nina Helander & Vilma
Vuori
14. Tunnistaako kone tunteesi?
Sävyanalyysi sosiaalisen median
sisältöjen tulkinnassa. Tuomo Helo
& Harri Jalonen
Available from: https://vastapaino.fi/sivu/tuote/twitter-viestintana/2442557
15. www.hamk.fi
Sentiment analysis of tweets about IBM
21.2.2019
Source:
Dejan Trifunovic 2019
Business Analytics and Business Intelligence
16. www.hamk.fi
Twitter sentiment versus Gallup Poll of
Consumer Confidence
Source: Brendan O'Connor, Ramnath Balasubramanyan, Bryan R. Routledge, and Noah A. Smith.
2010. From Tweets to Polls: Linking Text Sentiment to Public Opinion Time Series. In ICWSM-2010
18. www.hamk.fi
Communication styles in Twitter
• Energy sector and climate
change related Twitter
discussions were analyzed
• Who, and what kind of
communication styles were
found?
• Communication styles of tweets
• Source: Ketonen-Oksi & Jalonen
2017