IS3183

Management & Social Media

Lecture 5

Ricky FM Law
Key Concepts from Last Lecture
 Four Types of Network E ects

Positive Indirect Cross Side E ect

Positive Direct Same Side E ect

Example: VISA Card Payment /

Example: Phone Company

Trip Advisor

Negative Direct Same Side E ect

Negative Indirect Cross Side E ect

Example: Dating/Matching
Example: UBER
Services
ff
ff
ff
ff
ff
Watch Chapter 4 Video Again
 How are Network Effects Related to
Types of Social Media Platforms?

• Three types of social media

• Four types of network e ects

• Networking
• Direct, same side

• Content
• +ve and -ve

• Service • Indirect, cross side

• +ve and -ve

ff
SWAT Practice for Final Exam
 • Tip 1

• Crack open the question =

determine the number of
parts

• Tip 2

• Time management = allocate

time to speci c action /
content parts

• Your time allocation has

nothing to do with your
strength

https://www.amazon.com/Exam-Tactics-Chris-Prior-ebook/dp/B0765ZXHD7
fi
Tip 3

• Do an outline

http://www.outlineproductions.co.uk/
How are the design principles of modularity and
layering connected to end-to-end architecture?

• Do an outline for your

answer.

Follow this format (Assume 60 minutes)

1) Item one, xx minutes

2) Item two, yy minutes

3) …
 How are the design principles of modularity and
layering connected to end-to-end architecture?

1. Read the question several times, gure how to answer it and layout down the
outline — 5 min

2. De ne design principles of modularity — 5 min

3. De ne design principles of laying — 5 min

4. Explain end-to-end architecture — 10 min

nce
fere
5. Explain how 2 and 3 connect to 4 — 25 min
a re
nl y
i s iso
6. Rest and cool o for one minute
Th
7. Review and edit— 10 min

• Organisation / Argument logic / Accuracy / Style / Grammar

fi
fi
ff
fi
Chapter 5
The Structure of Platform Participations
• User Participations

• Following, Liking, Commenting, Posting …

• Engine of data production

• Purpose, to capture

• Users’ tastes, consumption habits, political orientations

• Sold to

• Analytic companies, data broker, advertisers

 Of ine vs Online Behaviours

• We chat, book a restaurant, listen and share a song both

online and o ine

• For o ine activities, nobody checks and gathers data.

Social Media do.

• For o ine, we do not follow (stalk) a friend. We do on

social media.
fl
ffl
ffl
ffl
 Learning Outcomes
• de ne user-platform participation in terms of platform design
and implementation

• explain how encoding of user participation works

• de ne social data and relate them to platform participation

and platform strategy

• distinguish di erent kinds of data produced by social media

• de ne core and peripheral interaction

• elaborate on di erent types of designing and implementing

core and peripheral interactions and the strategies social
media platforms use to accomplish these objectives.
fi
fi
fi
ff
ff
• Social media platforms are

• economic intermediaries,

• social and communicative arrangements and

• complex technical architectures of modules and

functionalities

• Shape the terms on the basis of which users get involved

and interact with other users on the platform to serve the
objectives they pursue as business organisations.

• In other words, they frame users’ behaviours to generate

data for business gains
• User participation on social media is not spontaneous.

• Each of the activities users carry out on social media is

mediated by a careful design that re ects the business
objectives of social media platforms and the ways these
objectives can be accommodated by technological
functionalities or a ordances.
ff
fl
 How do Social Media Transform
Platform Participation into Data Streams

• To produce social data, the platform needs to program

activities that are both easy to master and interesting
enough to be repeatedly and recursively enacted by very
large numbers of users.

• The encoding of platform participation into data follows a

very simple structure. As indicated, participation is
disassembled into easy-to-do actions that, thanks to how
easy they are to perform, facilitate userengagement and
participation and are constantly repeated.
• Social media platforms are data-driven organisations that
use platform participation as the principal means for
producing a variety of data on people’s behaviour.

• Such data can be used to tailor advertisements and other

platform interventions of a commercial or operational (to
improve platform functioning) nature.
• Platform participation recreates online a social context
whereby a set of well-de ned interactions can be
performed by users and directly encoded into data by the
platform’s system. For instance, following, sharing,
tagging …

• Social data are produced through the encoding of online

interaction and user platform participation.
fi
 • On social media, everything is de ned as ‘objects’. Users are
de ned as objects, as are videos, comments, photos and so forth.

• For instance, when a user follows another user, two user objects

are connected; similarly, when a user likes a photo these two
objects become connected.
fi
fi
• Social data are thus ready to be counted, correlated,
stored and eventually reutilised by third parties in a
number of ways.

• This is the power of social media encoding: it makes

visible, standardised and countable social interactions
that were invisible (perhaps even non-existent), variable
and not amenable to calculation before.
 Encoding

• Encoding of social media (Alaimo and Kallinikos 2016,

2017).

• Social media formalise a di use and largely informal

social ‘everyday’ into a set of platform activities that
produce social data whenever they are performed.
ff
Watch Chapter 5 Video on VLE
 How Facebook Tracks Your Data

https://www.youtube.com/watch?v=JAO_3EvD3DY
 What Separates ‘buy’ on Amazon and
‘like’ on Facebook
• For Amazon “buy”, a user completes a transaction.

• A Facebook “like” is no more than an expression of an opinion

or sentiment. An action of this type generates a di erent type of
data than simply recording a fact (a transaction).

• Opinion, sentiments, comments (even friendship) are ambiguous

actions and all highly dependent on the context in which they
occur. It matters, in other words, where a user is and who they
are with when they express a belief, an idea, a preference.

• Accordingly, how user participation is encoded into data is a

much more complex and delicate task than is the case when
recording ordinary transactions (e.g. buying an item).
ff
Uniqueness of Social Data

• Transaction data usually records an exchange

• Tracking data produced by cookies and tracking devices

(such as beacons), record simple online behaviours (time
spent on websites, searching habits etc.)

• Social data are produced by social activities carried out

on the platform by users. Activities of this sort are pre-
programmed by platforms in an online space that is
constantly tuned in and adjusted to user behaviour.
Platforms are purposely designed to let users participate
in speci c ways.
fi
 Social Data vs Online/Transactional Data

• Online Tracing / • Social Data

Transactional Data

• Opinions, emotions,
• Speci c action - Buying sentiments

on Amazon

• Intentions
• Cookies/Beacons to
track online behaviours
fi
 Twitter as an Example:
Core vs Peripheral Interactions

• On Twitter, users will only see the tweet stream of the users they
follow, ordered by the Twitter’s algorithm on the basis of the most
relevant tweet. What users tweet or retweet is based on what users
see; this in turn is based on algorithmic, ltering mechanics and user
following data.

• Tweeting is the core interaction of Twitter. The tweeting action

produces di erent data. It produces social data from the action of
tweeting, content in the form of text and location data (if the location
function of the device in use is turned on).

• Additionally, a single tweet may produce content data of another

kind, if users upload a picture or link an article. Even more
interestingly, a single tweet may embed other actions such as, for
instance, a user’s mentioning of other users or a hashtag, which we
call peripheral interactions (mentioning, hashtagging, etc.).
ff
fi

Different Types of Data

User Generated Content (UGC)

• It entails the creation and subsequent posting or uploading of

content such as video, text-based comments or views, photos.
User platform participation evolves, in fact, around user-generated
content and the speci c actions we have been describing as
encoding – that is, viewing, commenting, uploading, posting,
tagging, liking and so forth.

• UGC provides high volumes of data, such as images, posts of

various kinds and written comments that, being in the format of
text, are usually called unstructured.

• User-generated data are more di cult to compute and correlate

by using the current database techniques and available
technologies.
fi
ffi
• However, UGC provides the means and the context by
which platform participation can be performed and
encoded as social data.

• It is also very important to point out that when users act

on a piece of content – for instance, when users like a
photo – they link that content to structured data which
make user-generated content computable as a reference
object.
• For instance, when ‘Cristina likes a video’, the action of
liking produces social data that qualify both the object
Cristina and the object video. In this sense, even if the
video is unstructured data, the structured data generated
by performing liking or tagging make it referable and
ultimately computable.

• A tag is a descriptive label (metadata), attached to the

video allowing it to be measured and related to other
videos with the same tag. In this sense, the content data
from unstructured data become semi-structured data.
• It is thus important to distinguish between the content, say, of an
upload or post (what users generate as content) and the act of
uploading or posting that content (social data)

• As distinct from the content itself, posting, uploading, tagging

and so forth have signi cant value of their own. They indicate for
platform owners the preferences and choices of users – for
instance, what types of videos or photos they upload, what
people or content they are tagged to, who they interact with and
so forth.

• Users’ choices, therefore, expressed in acts such as posting,

uploading, viewing or liking, provide structured data (that is,
discrete clicks) that can be stored, counted, aggregated,
computed or scored to produce a range of user pro les and
chart the patterns of their platform participation.
fi
fi
 Value Creation
• Social data are structured data and, for this reason, have
a strong potential value for platform owners, advertisers
and third parties such as data analytics companies.

• Social media platforms are data-driven organisations that

use platform participation as the principal means for
producing a variety of data on people’s behaviour.

• Such data can be used to tailor advertisements and other

platform interventions of a commercial or operational (to
improve platform functioning) nature.
 Types of Data

• Structured data

• Descriptive data - pro le

• Behaviour data - click, follow, like

• Unstructured data

• User Generated Content

• Behaviour data structure UGC data, making it relatable

and computable
fi
 What is the Difference
between Structured and
Unstructured Data?

https://www.youtube.com/watch?v=MjjtfDsPmHA
https://businessoverbroadway.com/wp-content/uploads/2014/07/Data-De nition-Framework.png
fi
 UGC

• Predominantly unstructured, usually cast in the form of

text – verbal narration such as stories, opinions, blogs on
politics – or in the form of images such as photos and
videos.

• Clearly, stories cannot easily be added or multiplied,

subtracted or divided in the same way as, say, ratings of
hotels or places, Facebook’s likes or structured
information about individuals obtained by following
standard CV templates
 Text Analytics / Sentiment Analysis

• Text Analytics or Text Data Mining involves the

categorising and clustering of text, concept extraction
and word frequency measurements. These techniques
basically transform the open nature of texts into distinct
elds that are registered into a database that can be
ploughed by a machine and computed or otherwise
ltered through key words and other data queries.

• Sentiment analysis is another similar technique of parsing

unstructured data with the view to identifying attitudes,
values and opinions expressed in the form of texts.
fi
fi
 • The functional utility and achievement of both computer-
based text and sentiment analytics remain questionable
both in terms of validity (what they measure) and relevance
(how useful is what they measure). Furthermore, text and
sentiment analytics require considerable specialised
human resources to be carried out successfully.

• It is thus preferable to de ne to ensure a steady ow of

machine-readable, good-quality structured data.

• This is exactly what the identi cation of the core

interaction of a platform achieves, together with the
speci cation of the routes along which platform
participants interact.
fi
fi
fi
fl
Core & Peripheral Interactions

• Social media platforms design their core interaction

based on the business area in which they operate.

• For instance, networking-based social media design their

platform participation around the core interaction of
friending or following

• But the core interaction of a social media platform is

always complemented by peripheral interactions such as:
commenting, posting, liking.
• In fact, every social media platform, irrespectively of its
type (networking-, content- or service-based), rests on a
core interaction complemented by a set of peripheral
interactions that together create the suitable blend of
actions required to sustain and promote user platform
engagement.
• It is important to distinguish between the content, say, of
an upload or post (what users generate as content) and
the act of uploading or posting that content (social data).
 Twitter as an Example

• Tweeting is the core interaction of the platform. A single

tweet can produce social data (the act of tweeting), sensor
data (i.e. location), content data (i.e. the text of the tweet),
but it can further embed other actions — peripheral
interactions, like mentioning, hashtagging, or linking.
Furthermore, another user viewing that same tweet may like,
retweet or mention it. All of these actions, and the social data
they procure, are in some sense triggered by the action of
tweeting.

• Tweeting is the action that produces social data as a granular

and countable instance of user behaviour on the platform. It
is the core activity that the platform promotes, but tweeting
is also the activity that somehow triggers all the others.
• Twitter, a content-based platforms spreading brief text-
based content.

• Similar to Twitter, Instagram operates with a core interaction

which triggers a complex pattern of platform participation.

• On Instagram, the main activity is certainly uploading

photos, the core interaction of uploading procures both
social data and content.

• As tweettexts are the main content on Twitter, insta-photos

are the main content on Instagram. However, following and
liking are important for the platform functioning and
business model.
• For service-based platforms, the core interaction is usually
designed as collaboration or collective creation or sharing of a
service that may derive from knowledge or information, a good
that may be leased out or collectively shared, a skill or even time.

• TripAdvisor orchestrates its platform functioning and its entire

business model around the core interaction of user rating and
reviewing. Peripheral interactions on TripAdvisor have grown
immensely over time and now span from viewing to liking and
comparing to booking.

• As these observations indicate, platform participation, cast in

terms of both core and peripheral interactions, has become a
central element in supporting innovation in the travelling sector
and bringing forward an entirely new organising logic of travelling,
triggered by the participation of the crowd.
 Sum Up
• Social media are based on the production of social data.
The ways they design platform participation is integral to
this data strategy. The encoding of platform participation
happens along simple yet e ective routes.

• Participation is decomposed into action-data connecting

objects. The data thus produced open a huge spectrum
of opportunities for data use and recombination. Patterns
can be detected that reveal user consumption habits,
political orientations, opinions and tastes.
ff
 • Platform participation is designed for data (platform
functioning) and business purposes. Certain platforms
design a speci c set of actions because of the data they
procure (Instagram); others start from a core interaction in a
speci c business sector to develop a host of subordinated
interactions that relate to the provision of services
(TripAdvisor). Some platforms remain generally oriented in
supporting a core networking interaction (Facebook).

• It is important to understand how platforms design

participation, what data they produce and for what reasons.
Only in this way it is possible to obtain a fuller appreciation
of the waves of innovation social media bring to the
economy and to society.
fi
fi
 Can you …
• explain the link between social participation and social
data

• explain what encoding means and how it operates

• make a distinction between core and peripheral

interaction

• explain the di erent kinds of data produced by platform

participation

• distinguish between structured, semi-structured and

unstructured data.
ff
Test your Knowledge and Understanding

1. In what ways are user-generated content and social data

essential in sustaining social media platforms and the
services they provide?

2. What is the di erence between user-generated content

and social data?

3. Based on the readings suggested, explain the process of

encoding of a social media platform of your choice,
making a distinction between core and peripheral
interaction.
ff
 $$
Pro ling, Predictive
Marketing

Data Capture
Big Data
m nt Social
or
l a tf me Media
P ove Platform
p r
Im
Engine (Tech)

Social Data Behavioural User Generated Content

• Computable Encode
Descriptive
Data - Likes, (UGC) — Uploads …
• Portable Data - Pro le
Share videos, texts, photos
• Combinable
Structured Data Unstructured Data

Convert unstructured data into structured data

through Behavioural Data, making it relatable
and computable

Graphic Copyright: ricky fm law; Content from “Encoding the everyday: the infrastructural apparatus of social media”
fi

fi

Essential Readings
See you Next Week