SThAR_corporate_presentation_19-09-2009

A new model to predict social changes…

Confidential and Copyrighted2

What these events have in common ? Electrical & telephone bills, stock prices, frauds & deaths rates

The Percolation theory and the nuclear multi-fragmentation

The abundances of genes in various organisms and tissues

The churn distribution in a mobile network operator

The density distribution of votes in political elections

The density distribution of urban agglomerations

The distribution of firm-sizes all over the world

The frequency of words in natural languages

The scientific collaboration network

The total number of cites of physics

The Linux packages links

The Internet traffic…

… Are social events determined by universal laws ?


Unexplained Incredible Facts

Physicist John Archibald Wheeler stated that: “All things physical are information-theoretic in origin and this is a participatory universe...”

Regularity in the words frequency in natural languages and urban agglomeration was empirically already observed, about 100 years ago: The Zipf’s Law.

Benford’s Law provides results which have been found to apply to street addresses, population numbers, lengths of rivers, physical and mathematical constants, and processes described by power laws (which are very common in nature).


A little bit of History… background 75 years ago, George Kingsly Zipf (1902-1950), an American linguist and

philologist who studied statistical occurrences in different languages, observed incredible regularities, not only in the different languages but in several social distributions. Zipf was Chairman of the German Department and University Lecturer at Harvard University. He worked with Chinese languages and demographics as well.

Zipf's law states that while only a few words are used very often, many or

most are used rarely, following a proportionality of this type: Pn ~ 1/na,

where Pn is the frequency of a word ranked nth and the exponent a is almost 1. This means that the second item occurs approximately 1/2 as often as the first, and the third item 1/3 as often as the first, and so on.

The rank vs. frequency distribution of individual incomes in a unified nation approximates this law. Breaks in this "normal curve of income distribution" portend social pressure for change, even revolution. This is demonstrated in his 1941 book, "National Unity and Disunity" in which the break in the curve of income distribution in 1940 in Indonesia predicts revolution there. Revolution began five years later, in 1945.

.


Zipf’s Law

Zipf's law, an empirical law formulated using mathematical statistics, refers to the fact that many types of data studied in the physical and social sciences can be approximated with a Zipfian distribution, one of a family of related discrete power law probability distributions.

Up to now, it was not known why Zipf's law holds for most languages. However, it may be partially explained by the statistical analysis of randomly-generated texts.

If the natural log of some data are normally distributed, the data follow the log-normal distribution. This distribution is useful when random influences have an effect that is multiplicative rather than additive. .

Much of his effort could explain properties of the Internet , distribution of income within nations, and many other collections of data

http://en.wikipedia.org/wiki/Natural_log

http://en.wikipedia.org/wiki/Normal_distribution

http://en.wikipedia.org/wiki/Log-normal_distribution




Benford’s Law Benford's law, also called the first-digit law, states that in lists of numbers

from many (but not all) real-life sources of data, the leading digit is distributed in a specific, non-uniform way.

According to this law, the first digit is 1 almost one third of the time, and larger digits occur as the leading digit with lower and lower frequency, to the point where 9 as a first digit occurs less than one time in twenty. This distribution of first digits arises logically whenever a set of values is distributed logarithmically.

Real-world measurements are often distributed logarithmically (or equally, the logarithm of the measurements is distributed uniformly).

A logarithmic scale bar. Picking a random x position on this number line, roughly 30% of the

time the first digit of the number will be 1 (the widest band of each power of ten).

http://upload.wikimedia.org/wikipedia/commons/6/68/Logscale.svg

http://en.wikipedia.org/wiki/Logarithmic_scale


Benford’s Law This counter-intuitive result has been found to apply to a wide variety of data

sets. The results also hold regardless of the base in which the numbers are expressed, although the exact proportions change.

It has been argued that Benford's law is a special case of Zipf's law. This special connection between these two laws can be explained by the fact that they both originate from the same scale invariant functional relation from statistical physics and critical phenomena

http://en.wikipedia.org/wiki/Benford's_law




In mathematical statistics and information theory, the Fisher Information (sometimes simply called Information) is the variance of the score. Its role in the asymptotic theory of maximum-likelihood estimation was emphasized by statistician R.A. Fisher

The Fisher information is a way of measuring the amount of information that an observable random variable X carries about an unknown parameter θ upon which the likelihood function of θ, L(θ) = f(X;θ), depends.

Fisher information is widely used in optimal experimental design. Because of the reciprocity of estimator-variance and Fisher information, minimizing the variance corresponds to maximizing the information.

When the linear statistical model has several parameters, the mean of the parameter-estimator is a vector and its variance is a matrix. The inverse matrix of the variance-matrix is called the "information matrix". Using statistical theory, statisticians compress the information-matrix using real-valued summary statistics; being real-valued functions, these "information criteria" can be maximized.

Fisher Information


Dunbar's number is a theoretical cognitive limit to the number of people with whom one can maintain stable social relationships. These are relationships in which an individual knows who each person is, and how each person relates to every other person.[1] Proponents assert that numbers larger than this generally require more restricted rules, laws, and enforced norms to maintain a stable, cohesive group. No precise value has been

proposed for Dunbar's number, but a commonly cited approximation is 150.

Dunbar's number was first proposed by British anthropologist Robin Dunbar, who theorized that "this limit is a direct function of relative neocortex size, and that this in turn limits group size ... the limit imposed by neocortical processing capacity is simply on the number of individuals with whom a stable inter-personal relationship can be maintained." On the periphery, the number also includes past colleagues such as high school friends with whom a person would want to reacquaint themselves if they met again

Dunbar’s number

http://en.wikipedia.org/wiki/Interpersonal_relationship

http://en.wikipedia.org/wiki/Social_group

http://en.wikipedia.org/wiki/United_Kingdom

http://en.wikipedia.org/wiki/Robin_Dunbar

http://en.wikipedia.org/wiki/Neocortex


Six degrees of separation (also referred to as the "Human Web") refers to the idea that, if a person is one step away from each person they know and two steps away from each person who is known by one of the people they know, then everyone is at most six steps away from any other person on Earth. It was popularised by a play written by John Guare.

Kevin Bacon Game: The game "Six Degrees of Kevin Bacon" was invented as a play on the concept: the goal is to link any actor to Kevin Bacon through no more than six connections, where two actors are connected if they have appeared in a movie together.

Six degrees of separation

http://upload.wikimedia.org/wikipedia/commons/9/94/Six_degrees_of_separation.png

http://en.wikipedia.org/wiki/Six_Degrees_of_Separation_(play)

http://en.wikipedia.org/wiki/John_Guare

http://en.wikipedia.org/wiki/John_Guare

http://en.wikipedia.org/wiki/Six_Degrees_of_Kevin_Bacon

http://en.wikipedia.org/wiki/Six_Degrees_of_Kevin_Bacon

http://en.wikipedia.org/wiki/Kevin_Bacon

Confidential and Copyrighted


OBSERVATION

• 2008 SpanishElections

INTUITION

• Search for similar empirical cases responding to samerules => Brazilelections, Populationdistribution….

MODELIZATION

• Empirical simulationsexhaustive analysisto identify theparameters and pattern behind the : RULE OF UNIVERSAL SCALE

UNIVERSAL LAW

• Looking for a Theoreticalframework, based on existing laws and sicentific principles

Background: How it all started ?

¡

• Based on a classical scientific process, a scientist’s curiosity him led to analyze the results of the 2008 Spanish elections and …


OBSERVATION


INTUITION


MODELIZATION


UNIVERSAL LAW


Background

• Based on a classical scientific process, a scientist’s curiosity led him to analyze the results of the 2008 Spanish elections and discovered that their logarithmic representation followed , surprisingly, almost a perfect straight line.

Bipartisanship

Other parties with representation

Parties with no representation

Ln (

num

ber

of

vote

s)

Ranking


OBSERVATION


INTUITION


MODELIZATION


UNIVERSAL LAW


Background

• Based on a classical scientific process, a scientist’s curiosity led him to analyze the results of the 2008 Spanish elections and discovered that their logarithmic representation followed , surprisingly, almost a perfect straight line.

• Such regularity drove the scientists to search for a pattern behind that behavior, and to explain how people associate to create common interests groups (clusters).


OBSERVATION


INTUITION


MODELIZATION


UNIVERSAL LAW


Background

• Expecting this behavior could be exhibited in other real life events and searching for possible analogies, similar results were found in several interesting data collections…

• Like in the result of the Brazilian elections…

http://images.google.es/imgres?imgurl=http://api.ning.com/files/GZxJ6Kr9w-vTyIFBYQcUNytLoub6ac8F04jq-4Cx5BlwuL5zHb8M*5MLCgsk5KGbTkvLTJCsf2NDmh-hlNoPYM3n2aq4u1X5/Brasil.jpg&imgrefurl=http://omlatam.ning.com/profile/JuanCassa&usg=__1m8ij2KgRGKfg6rERbAlCRjnTrY=&h=560&w=800&sz=24&hl=es&start=8&um=1&tbnid=Np2EYjcbqi8uyM:&tbnh=100&tbnw=143&prev=/images?q=brasil&hl=es&um=1


OBSERVATION


INTUITION


MODELIZATION


UNIVERSAL LAW


Background


• Like in the result of the Brazilian elections…

• and in the population distribution in towns in different countries !!…

http://images.google.es/imgres?imgurl=http://api.ning.com/files/GZxJ6Kr9w-vTyIFBYQcUNytLoub6ac8F04jq-4Cx5BlwuL5zHb8M*5MLCgsk5KGbTkvLTJCsf2NDmh-hlNoPYM3n2aq4u1X5/Brasil.jpg&imgrefurl=http://omlatam.ning.com/profile/JuanCassa&usg=__1m8ij2KgRGKfg6rERbAlCRjnTrY=&h=560&w=800&sz=24&hl=es&start=8&um=1&tbnid=Np2EYjcbqi8uyM:&tbnh=100&tbnw=143&prev=/images?q=brasil&hl=es&um=1

http://images.google.es/imgres?imgurl=http://www.iglesia-sanpablo.org/flags/usa.gif&imgrefurl=http://www.iglesia-sanpablo.org/&usg=__gpf2EiNDguzCf4FsHQuZzSrBJsE=&h=302&w=575&sz=9&hl=es&start=4&um=1&tbnid=XvivmtNQv1ecZM:&tbnh=70&tbnw=134&prev=/images?q=usa&hl=es&um=1

http://images.google.es/imgres?imgurl=http://www.jobinitaly.biz/english/images/articles/Flag of italy.gif&imgrefurl=http://www.jobinitaly.biz/english/article_italian-flag&usg=__BlEnTuPEb2pMLwqcYSwgWFTbZt8=&h=265&w=390&sz=5&hl=es&start=1&um=1&tbnid=Lp68aftPtQzdtM:&tbnh=84&tbnw=123&prev=/images?q=italy+flag&hl=es&um=1

http://images.google.es/imgres?imgurl=http://nuke.conditioningzone.com/Portals/0/Flags/German flag.JPG&imgrefurl=http://nuke.conditioningzone.com/&usg=__16CPO4b9B00gbhxZlrbYFmb401E=&h=366&w=606&sz=6&hl=es&start=2&tbnid=X_KmT0Z10cBvCM:&tbnh=82&tbnw=136&prev=/images?q=german+flag&gbv=2&hl=es


OBSERVATION


INTUITION


MODELIZATION


UNIVERSAL LAW


Background: Nothing New Yet!


• Like in the result of the Brazilian elections……

• and in the population distribution in towns in different countries !!…

In fact, this regularity is what The Zipf’s Law had confirmed for many years.


OBSERVATION


INTUITION


MODELIZATION


UNIVERSAL LAW


Background

• Similar regularities have been detected in many other social events…

BUT …

NOBODY HAD BEEN ABLE TO EXPLAIN HOW or WHY THESE REGULARITIES HAPPENED BASED ON SCIENTIFIC PRINCIPLES…

…AND THERE WAS NEVER A THEORY AVAILABLE TO EXPLAIN IT !!

Up to now !! ...

Zipf’s Law

•Distribution of words•Distribution of firm sizes•Internet traffic distribution…

Benford’s Law

•Deaths rate•Population number•Stock Prices…


OBSERVATION


INTUITION


MODELING

• Empirical simulationsexhaustive analysisto identify theparameters and pattern behind this

UNIVERSAL LAW


Scientific Innovation

• Attempting to define a scientifically based framework, which would explain and predict the nature of those events, after trying different models and methodologies, their first conclusion (based on a sound scientific empirical analysis) led to the discovery of a new completely revolutionary scientific concept…

… The UNIVERSAL SCALE RULE… !!

UNIVERSAL SCALERULE


OBSERVATION


INTUITION


MODELING


UNIVERSAL LAW



• Thanks to the application of this rule, it was proven that the way people form groups of interest (clusters) follows a pattern based on a thermodynamic variable that was named “COMPETITIVENESS” (λ)

The pattern behind it all had just been unveiled !!

UNIVERSAL SCALERULE


OBSERVATION


INTUITION


MODELING


UNIVERSAL LAW



• It was the first time that the Information Theory was successfully used to explain reality and the first time that from the “Principle of Minimization of Fisher Information” could be obtained, not only the Relativity, Quantum Mechanics, Classical Electrodynamics, Field Theory, Thermodynamics... but, as well, an innovative theory:

“Social Thermodynamics”

InformationTheory

Relativity Theory

Quantum Mechanics

Thermodynamics

SOCIAL

THERMODYNAMICS


OBSERVATION


INTUITION


MODELING


UNIVERSAL LAW

• Looking for a Theoreticalframework, based on existing laws and sicentific principles, theZipf’s Law i(social) s shown to be as universal as the Gas Law (physical)


1. Explains both Zipf’s Law and Benford’s Law: they naturally emerge when the correct symmetry and variables are introduced in the Information Principle (“Zipf's Law from a Fisher variational-principle”, http://arxiv.org/abs/0908.0501).

2. Confirms the analogy between the properties of Social Systems and the Thermodynamics of Gases and Liquids through the “Scale-Free Ideal Gas” (SFIG):

Therefore, the Zipf’s Law is so universal as the Universal Gas Law, as they rise from the same principle, but with different symmetries (“Fisher-information and the thermodynamics of scale-invariant systems”, respectively, http://arxiv.org/abs/0908.0504).

SOCIAL

THERMODYNAMICS

UNIVERSAL LAWS

• Zipf’s Law

• Benford’s Law

• Dunbar’s Number

• Six degrees of separation

• “Competitiveness”

http://arxiv.org/abs/0908.0501



OBSERVATION


INTUITION


MODELING


UNIVERSAL LAWS

• Zipf’s Law

• Benford’s Law





3. STh applied to the Network Theory (“Unravelling the size distribution of social groups”, http://arxiv.org/abs/0905.3704) explains classical” predictions:

Prediction of the max. number of contacts (connections) a human can keep, widely known as the "Dunbar's number”

Prediction of the min. average distance between any 2 people on the Earth, known as the “Six degrees of Separation”, is a conseq. of Dunbar’s Number.

UNIVERSAL LAWS

• Zipf’s Law

• Benford’s Law




SOCIAL

THERMODYNAMICS

Thanks to Social Thermodynamics, those properties that had been previously detected, only empirically, now have a valid scientific explanation and a model which can be applied to any event.



OBSERVATION


INTUITION


MODELING


UNIVERSAL LAW

• COMPETITIVENESS: Thenew variable that makespossible to unveli thepattern behind empiricalevidences and get a universal law


• Social Thermodynamics is not only able to explain classical known properties, but also to predict and detect patterns that have not been discovered yet, such as:

• E.g. Prediction of the pattern behind the ‘City-Size Distributions’ and ‘Electoral Results’.

The discovery of a new ‘Universal Scale Rule’ leads to the definition of “Competitiveness”, a new thermodynamic variable that allows to classify and simulate the way people join to create groups of common interests. The way these groups are created and distributed depends totally on the Competitiveness parameter

SOCIAL

THERMODYNAMICS

•The pattern behind the real natural events and its results has been now unveiled and explained based on a Universal Law that can be applied to predict other social patterns.



OBSERVATION


INTUITION


MODELING


UNIVERSAL LAW

• Looking for a Theoreticalframework, based on existing laws and sicentific principles, theZipf’s Law i(social) s shown to be as universal as the Gas Law (physical)

SOCIAL

THERMODYNAMICS

UNIVERSAL LAWS

• Zipf’s Law

• Benford’s Law




As regularity exists, and the theoretic framework has been created, results can be modeled and, therefore, predicted !


Applications• Understand the Network Structure, Clusters Distribution…

• Identify first “churners”: Thermodynamically Ebullition Points

• That has an enormous impact in many ways:

• Predict with unrivalled accuracy future churners to retain them

• Quickly spread patterns/information in a costly-efective way

• Fight against threats making use of these ebullition points

• E.g.: Viruses spreading on Data/Mobile networks

From the Network Analysis

• Make Simulations, predicting the dissemination of opinion or theNetwork robustness in case some elements are lost.

• Test the best Dissemination Strategy and Detect the WeakestPoints in the network.

From the Computationalrecreation of the Network

• Network = Thermodynamical conditions dependant material :

• Some favorable conditions will favor the growth of the networkand its robustness

• Other conditions may cause its disappearance/weakening..

• Discover these conditions and predict the effect of a change in these conditions

From the Analogy withThermodynamics


Benefits applied to organizations• By modeling their users/customers patterns they can

better understand their habits/behaviours and actconsequently

BETTER KNOWLEDGE OF THEIR CUSTOMERS/USERS

• Get the power to classify their users base in differentsegmentation clusters based on simple parameters.

POWERFUL SEGMENTATION TOOLS

• Predict interactions and advance future changes(opportunities and threats) in their social members

PREDICTABILITY

• Dramatically improve the communication/promotionalactions/campaigns by filtering only the target segments/users, resulting in a more efficient marketing

OPTIMIZE MARKETING AND COMMUNICATION COSTS

• Be able to play focusing on identified potentialchurners or in key initial viral points instead of spreading efforts throughout the entire base.

IDENTIFY “EBULLITION POINTS” & “ACHILLES HEELS”


Solution areas• Marketing

• Viral Marketing Optimization , Media Investment Optimization, Social Networks…

• Threats & Churn Detection• Disruptive Churn Analysis (vs statiscal existing models)

• Telco Networks (e.g.: Network viruses spread)

• Enterprises facing Web 2.0 new malware

• Fraud Prevention• Healthcare Fraud

• Insurance Fraud

Commercial Applications

• Security, Intelligence and Defense• Counterterrorism, Investigation (Crime, Narcotraffic, Money Laundering…

• Electronic Crime: Online/Computing/Mobile/Email/I.M. Analysis …

• Geo/Social/Political Analysis• Social/Demographical Analysis, Population Distribution, Social Trends..

• Electoral Analysis, Predictions and Fraud detection

• Research• Research Public Institutions, Transnational Projects…

• Knowledge Management and Collaboration , NGOs…

Governments & Academic


Company Overview About SThAR

SThAR is the developer and world’s leader in the application of Social Thermodynamics Universal Laws to real business needs, which can provide revolutionary solutions, under a whole new scientific methodology.

Vision

Through the analysis of network properties and the use of recently discovered physical principles, SThAR will help public and private enterprises to predict and model social interactions for multiple purposes, delivering an unsuspected powerful tool to explain social events and gain a dramatic competitive advantage:

Identifying the network Ebullition Points and, therefore, the best and most cost-effective Dissemination Strategies (e.g., for viral marketing purposes).

Detecting the Achilles’ Heels and most risky areas susceptible to be threatened leading to the weakening/destruction of the network (for churn reduction, detection of mobile viruses spread, etc)

Anticipating social changes and predicting future results thanks to a new theoretical framework, rather than using traditional empirical approaches.


Why is SThAR revolutionary ?What makes SThAR unique:

A disruptive and whole new approach:

Versus more than 100 years of conventional empirical based models

With a mathematic model that, instead of using well known statistics methods, explains regularities through physical principles and can predict next ones.

Scientific foundation and recognition

SThAR scientists are the creators of the applied Social Thermodynamics .

Recognition of the scientific international community.

Correlating Operator’s and Environmental Networks

For a more comprehensive analysis, we can correlate Carrier’s data with social environment behavior, dramatically enhancing the power and accuracy of the combined prediction results.


Social Thermodynamics applicationMethodology

Project Methodology. 5 Phases:

1.DATA COLLECTION

2.NETWORK TYPIFICATION

3.ENVIRONMENT CROSS-RELATED

ANALYSIS

4.PROPAGATION SIMULATION

5.PREDICTIVE MODELING

AUTOMATION


Methodology

1. Data Collection (only 4 required fields)

DATA COLLECTION

• Timestamp

• Source

• Destination

• Duration


Methodology

2. Network Typification

SOCIAL NETWORK ANALYSIS

• Topology

• Network Properties

• Patterns Identification

• Measure of COMPETITIVENESS and Clusters Analysis

• Position and Size Distribution

• Inter/Intra connectivity

• Ebullition Points Detection

• Number and Strength of contacts and connections

• Centrality, Connectivity, Clustering Coefficient

• Propagation/Influence

• Achilles Heels Detection

• Identification of weakest points to reinforceprotection

• Network Resistance to nodes dissapearance

• Regeneration capacity


Methodology

3. Environmental Info Correlation

ENVIRONMENTAL CROSS-RELATED INFORMATION ANALYSIS

• Search of environment correlations to improvedissemination analysis.

• Thermodynamical properties identification of theunderlying social network


Methodology

4. Propagation and Network Behaviour Simulation

PROPAGATION SIMULATION

• Initial difussion nodes selection

• Analysis of neighbourhood infection power

• Multiple simulations changing parameters for fine-tunning

• Analysis of clusters behaviour: growing, melting, dissapearing…

• Identify best strategies to improve propagation process


Methodology

5. Predictive Modeling and Automation

PREDICTIVE MODELING AUTOMATION

• Identify the parameter of the model and formulate thetheoretical framework

• Eventually create and automated prediction tool basedon a few system inputs.

• Weather-like prediction tools


Application of Social ThermodynamicsBusiness Case: A Mobile Operator1 single model => 5 applications:

1. Disruptive Churn Analysis Model

2. Advanced Behavioral Marketing Clusters and roles identification and behaviors prediction

Thermodynamical conditions and environment variable correlations

3. Viral Marketing Optimization Media investment optimization by identifying the opinion makers/ebullition points…

4. New Promotions/Plans/Bundles Acceptance Simulation

5. Virus Quick Spread Prevention

5 Solutions for 5 main and totally different issues in a same company applying the same single Theoretical Model


Business Case: A Mobile OperatorMobile Network Recreation (NodeSnowball)

Call duration distribution

First Level Snowboard (Outcoming calls ) First Level Snowboard (SMS)

Color: Operator the user belongs to

Number of lines : Number of calls

Thicknes: Average call duration to that number


Degree centrality distribution in a

SFIN (Scale-Free Ideal Network)

Some examples of degree types

Business Case: A Mobile OperatorNetwork Modeling. Nodes classification

Analysis of network parameters (centrality, degree distribution, clustering…) allows to classify nodes(both, company and competitors nodes interacting each other) and identify the most influential/vulnerableones.That provides critical information about their behavioral pattern (influential power, propagationcapacity, and their relative position versus the interest groups and the global position in the wholenetwork).That classification is regardless other sociological variables (sex, ages, race…) that will only be used whendeciding how to communicate the messages in a Marketing campaign


Business Case: A Mobile OperatorMobile Network. ConnectionsStrength

Evolution of 2 contacts’connections weightsbased on a real mobileoperator bill.

Each peak representsan event. A high callfrequency helps tokeep high the value ofthe contact weightdespite the naturaldecrease in time.


Business Case: A Mobile OperatorMobile Network. SuscribersConnections

Illustration of a 1.000nodes network basedon the SFIN (Scale-FreeIdeal Network).

It is clearly visible theGiant component.


Business Case: A Mobile OperatorClusters identification

The modulation process allows to identify common interest groups , associate differentnodes into clusters, and hierarchically, clusteres into super clusters, based on localinteraction patterns, simplyfing network analysis at clusters levels instead of nodes level.


Business Case: A Mobile OperatorCorrelation with environment variables

Thermodynamic conditions identification

Thanks to Social Thermodynamics Theory, weknow now that the size distributions of thesegroups depends on a “Competitiveness” (λ)variable that defines the way humans formcommon interest groups (e.g.: if a given societygroups in many small clusters or in a few bigones, etc)

The value of λ , representative of every socialgroup, can be obtained from a number of availablesources (public databases, cities sizedistribution, electoral results, firms sizedistribution…)

That number (λ) has a definitive influence on thebehavior of the opinion transmission flows in asocial network, as people connections are the waythe information goes by.

Red: Empirical Data

Black: Social Thermodynamics Prediction


Business Case: A Mobile OperatorDramatically improving churn detection rates

•While existing Churn predictive models have an accuracy of no more than a 30%, theSocial Thermodynamics based methodology may dramatically increase this ratio.

•As an example, see the level of accuracy obtained by the Social ThermodynamicsMatematical model (black curve) compared to reality (red curve) vs other statiscalmodels (see green curve)


Business Case: A Mobile OperatorChurnAnalysis and Control

Thermodynamical analogy: CHURN = PHASE TRANSITION•A churner moving from one company to a competitor can becompared with a water molecule dropping out from ice to becomeliquid (melting).•Depending on local thermodynamical conditions(temperature, pressure, density…) some areas are more sensitive tophase changes than others.•Social Thermodynamics Theory provides the social equivalentframework to the classical Thermodynamics, making possible toanticipate the Suscriber’s Churn Sensitivity with the same accuracythat a phase transition can be predicted in the Condensed MatterPhysics, using an algorithm based on the last Montecarlodevelopments for Statiscal Physics.



Thermodynamical conditions•The churn process or phase transition depends on a newthermodynamical variable (εo) whose value can be obtained frompublic economical data, as it keeps relationship with the“Consumption Temperature”•It determines, in one side the Phase change probability distributionwhere ε1 is an specific system value and εo is an indicator of thecost/saving as a result of the transition.•In the other side, the “Economical temperature” allows todescribe, thanks to the Montecarlo method, spontaneous transitionsand massive acquisitions/lost of customers due to newpromotions/offers.



Potential Churners Identification:•The local description provides for every single node their socialthermodynamical status regarding phase transition; not only formembers in the nework but from competitor’s clients as well, as theyinteract with the first ones.•Thus, the algorithm allows to find out in the network the weakest areasand those ones with the highest growth potential, to optimizeinvestment in churn protection and new customers acquisition,protecting/increasing operator’s customer base•In addition, with this model it is possible to detect those nodes that canstart a chain reaction and predict the effects of a phase transition in thenetwork clusters generated, for example, by a competitive offer,selecting the areas where it will be more effective.



Potential Churners Identification:•The local description provides for every single node their socialthermodynamical status regarding phase transition; not only formembers in the nework but from competitor’s clients as well, as theyinteract with the first ones.•Thus, the algorithm allows to find out in the network the weakest areasand those ones with the highest growth potential, to optimizeinvestment in churn protection and new customers acquisition,protecting/increasing operator’s customer base•In addition, with this model it is possible to detect those nodes that canstart a chain reaction and predict the effects of a phase transition in thenetwork clusters generated, for example, by a competitive offer,selecting the areas where it will be more effective.


Churn Analysis:SThAR approach vs existing methodsExisting models are always

Statistics and based on 2 approaches

Usage pattern based: more complex and expensive, but more accurate

Billing: cost-effective, lower accuracy

Limited accuracy

Below 30% of accuracy in prediction (70% of churneres are not detected)

It is noted that amongst 30% customers, many of them will leave whatever actions telco takes. So most customers who are likely to defect will be gone!

Need to learn (neuronal learning) and be fed with many inputs

Look at themselves

In addition, we can provide a dynamic model including predictions about the health of the Competitors.

We don’t use Statistics to analyze empirically the PAST, but LAWS to predict the FUTURE


Business Case: A Mobile OperatorThe Power of Predicting Mktg. Effectiveness

New Promotions/Plans/Bundles Acceptance Simulation•While the “Economical Temperature” is a global variable, depending onthe general economical status, the specific ε1 is a local value that CANBE MANIPULATED through Marketing Campaigns.• Following the equivalence with classical Thermodynamics, ε0 wouldrepresent the average temperature in a room, while ε1 would representthe effect of turning on an oven or a refrigerator, i.e. the localmanipulation of the temperature.

•“Social temperature” can be modified and the result of that modificationcan be PREDICTED thanks to Social Thermodynamics•Those predictions can be used for a unlimited number of marketing simulations :testing the best Call rates plan, Flat fees, Bundles, Promotions…., having an unrivalled and unprecedented tool to gain a

definitive market competitive advantage.


Business Case: A Mobile OperatorAttacks and Rare Events Early Detection

Real Attack DetectionThe pattern followed by an SMS virus in a massive attack (e.g.) sending autonomouslymessages to contacts lists, will significatively differ from the usual human patterns, allowing toidentify suspicious behaviors. The follow-up of the “Tsunami’s front” when early detected willallow to identify the epidemy focus, obtain the properties of the infection dissemination andpredict its evolution to prevent or mitigate it

Rare events (Fast dissemination messages)Similar behaviour (though of a totally different nature) is observed in the fast dissemination ofnews/rumours/messages in very short periods of time (e.g. Xmas SMS, “pass-it now”messages…). It is important to differentiate the different types for many purposes:

• Detect and avoid, if necessary, the use of fraudulent messages (for legal compliance)• Empower users consumption to increase revenues• Avoid lines/network saturation in very shorts periods of time throug resourcesbalancing.


Example: Virus propagation simulation

Business Case: A Mobile OperatorVirus infection analysis

Fraction of network infected after 10 epidemies

randomly generated . Some epidemies don`t progress

while other extend quickly throughout the network.

Some of them don „t present any initial symptom but

suddenly start a wide spread of the infection.

Velocity of infection. Number of infected nodes

vs time. Generally, epidemies grow exponentially

until a maximum level of infection and then

decrease.



Fraction of network infected in 10

different simulations

The progress of the epidemy

(contagion potential) through the

network will highly depend on the

initial focus (“source of fire”)

Some starting points canwrongly be considered asnot dangerous, but thescenario can drasticallychange if sensitive nodesare reached, and theprocess become a massiveattack.

Generally the number ofinfected nodes growsgeometrically in earlystages to slow down when asaturation percentage ofinfected nodes is reached.


Nodes classification by their risk of contagion

Business Case: A Mobile OperatorVirus propagation simulation

Epidemy evolution in time (t = 1, 3, 5 & 7)Red : Infected nodes Green: Not infected nodes

Thanks to virus spread simulations we can:

•Obtain the disseminationpatterns• Classify nodes, i.e. dentifythose users susceptible to accelerate the process if they get infected (most contagious)and, therefore… • Design prevention/defenseplans.


Number of nodes infected in the network (total nodes:

1.000) vs Time based on the probaility of Infection (p)

Fraction of Network finally infected after the Virus

spread, based on the probability of infection (p)


To classify nodes by risk of contagion, a large number of dissemination processes are simulatedvarying the probability of infection. Depending on the time needed to reach saturation and the final fraction of network infected a value of infection power is assigned, representing their capacity togenerate and/or accelerate epidemies.



Graphical reproduction of the network showing the

first 10 levels of infected nodes, from the initial one.

Recreation of the infected component and the saved ones

after the Virus spread for p = 0,3


ASThROAutomated Socio Thermodynamics Research Operative

SThAR Marketing System, is a powerful all-in-one marketing solution that providesto Marketing professionals all the tools to create, plan, launch, track and automatedirect marketing campaigns including and advanced Content Manager System (CMS).

Based on the information obtained fromthe previous network analysis, modellingand predictive simulations, it allows toexecute , track and control differentonline marketing actions and generatethe corresponding reports to verify thesuccess of marketing campaigns , analyzedeviations and help to take correctivedecisions .


ASThROFeatures (I)

Subscribers TrackerYou will know all your subscribers’ actions through sent e-mail and web site

Campaign ManagerYou can define and planify online marketing campaigns for targetted emailing. Itincludes an approval workflow tool, preliminar emails preview, send scheduler…

Suscribers ManagerYou can import, export and create automatically suscribers lists, and add (OPT-in) or eliminate (OPT-out) customers at one-click

Segmented Suscribers ListYou can manage your suscribers based on segmentation predefinedcriteria, merge different lists into one, manage by users or by users’ profiles , etc.


ASThROFeatures (II)

Events and Response ManagementYou can control all the events related to any direct marketing campaign: sentmessages, read messages, bounced/failed messages… and analyze at one clickthe suscriber history , and have a granular control.

Real Time Redemption AnalysisYou can analyze the results of your campaigns in real-time, allowing to verify theeffectiveness of direct marketing promotions in launch time to quickly detectdeviations and take corrective actions.

On-line Analysis of Off-Line CampaignsYou can analyse the success of off-line marketing activities by geography, bymedia, by advertising supports (TV, press, radio…) through the inmediate off-line to on-line traslation.

Integration with Clients/External CRMsYou can export the results of the Campaigns Analyzer to Excel and other formatsto be used/integrated in external CRMs.


ASThROFeatures (III)

Client Profiles GeneratorAutomatic + Manual indicatorsAdvanced suscribers search, by any field/list/indicatorPossibility to add new indicators to preexisting suscribers/clustersPossibility to define individual preferences regarding different concepts: type of message, contact frequency, language, email format…

Business Oportunities DesignerYou can create your own business indicators to generate new oppotunities alerts.

Advanced Email DesignerYou can use the predefined templates and designs or use external ones. It allowsto manage Documents & Images to store in central servers all the componentsincluded in emails.

Advanced Newsletters DesignerYou can design, apply or modify existing templates for your periodical on-line marketing communications.


Business Case: A Mobile OperatorThe Definitive MarketingWeapon

For more information: [email protected]

www.sthar.com

What if reality was predictable ?

SThAR_corporate_presentation_19-09-2009

Business

zipfian distribution

empirical law

benfords law

social events

social distributions

frequency of words

words frequency

zipfs law states